Flownet

Added
- [425](https://github.com/equinor/flownet/pull/425) Added functionality to load production data from CSV file. Changed position of observation `vectors` entry in configuration file (now one level higher, same level as `simulation` and `database` in `data_source` entry).
- [417](https://github.com/equinor/flownet/pull/417) Added functionality to history match dissolved salts (TDS) in produced water.
- [404](https://github.com/equinor/flownet/pull/404) Added possibility for regional multipliers for permeability, porosity and bulkvolume multiplier. Current implementation allows for defining either one global multiplier, or a regional multipliers based on a region parameter extracted from an existing simulation model (typically FIPNUM, EQLNUM, SATNUM etc). The regional multiplier will be in addition to the per tube multipliers. New keys in config yaml are: porosity_regional_scheme (global, individual or regions_from_sim), porosity_regional (define prior same way as for other model parameters) and porosity_parameter_from_sim_model (name of region parameter in simulation model). The same three keys exists for permeability and bulkvolume_mult.
- [383](https://github.com/equinor/flownet/pull/383) Added option to either define a prior distribution for KRWMAX directly by using krwmax in the config yaml, or to let KRWMAX be calculated as KRWEND + delta. To do the latter, set krwmax_add_to_krwend to true, and then the prior distribution definition in the config yaml for krwmax will be interpreted as a prior distribution for the delta value to be added to KRWEND to get the KRWMAX.
- [386](https://github.com/equinor/flownet/pull/386) Expose FlowNet timeout to user.
- [356](https://github.com/equinor/flownet/pull/356) Added option to distribute the original volume over the FlowNet tubes in a Voronoi-diagram style. I.e., areas with a high density of FlowNet tubes get a lower volume per tube.
- [379](https://github.com/equinor/flownet/pull/379) Added option to let SWCR be calculated as SWL + delta, instead of providing the prior distribution for SWCR directly. To do this, set swcr_add_to_swl to true, and then the prior distribution definition in the config yaml for swcr will be interpreted as a prior distribution for the delta value to be added to SWL to get the SWCR.
- [372](https://github.com/equinor/flownet/pull/372) Added option to let the additional flownodes initially be placed within the original volume rather than within the convex hull of the real wells. To do this set place_nodes_in_volume_reservoir to true.

Fixes
- [403](https://github.com/equinor/flownet/pull/403) Fixes bug in generation of THPRES keyword for OPM Flow
- [391](https://github.com/equinor/flownet/pull/391) Fixes bug in generation of yaml files for visualization in Webviz.
- [372](https://github.com/equinor/flownet/pull/372) Fixes bug of hull_factor not actually being used for placing additional nodes outside the perforations.
- [374](https://github.com/equinor/flownet/pull/374) Fix for memory leak in result plotting script.

Changes
- [401](https://github.com/equinor/flownet/pull/401) When distributing volume from a simulation model to a FlowNet, the voronoi_per_tube distribution method now distributes first to tube cells, sums these values, and then redistributes the summed tube volume equally over the tube cells.
- [396](https://github.com/equinor/flownet/pull/396) When using scheme regions_from_sim for equilibrium regions, the user can supply which region to base the generation of FlowNet model's EQLNUM parameter on. The default region is EQLNUM, but other region parameters (such as FIPNUM or SATNUM) can be used by setting region_parameter_from_sim_model for equil in the config yaml file.
- [392](https://github.com/equinor/flownet/pull/392) When using scheme regions_from_sim for relative permeability, the user can supply which region to base the generation of FlowNet model's SATNUM parameter on. The default region is SATNUM, but other region parameters (such as FIPNUM or EQLNUM) can be used by setting region_parameter_from_sim_model for relative permeability in the config yaml file to e.g. EQLNUM.
- [383](https://github.com/equinor/flownet/pull/383) KRWMAX now defaulted to 1, but exposed to used. Previously it was hard coded to 1.
- [386](https://github.com/equinor/flownet/pull/386) Increase default timeout from 900 s to 3600 s.
- [363](https://github.com/equinor/flownet/pull/363) Drop Python 3.6 support.
- [361](https://github.com/equinor/flownet/pull/361) Adding numpy code for SWOF/SGOF table generation, instead of using pyscal, to improve computational efficiency when running many realizations with many SATNUM regions. Keeping pyscal in the tests for comparison.

Added
- [365](https://github.com/equinor/flownet/pull/365) Analytics workflow now also available for prediction mode.
- [351](https://github.com/equinor/flownet/pull/351) Added simple plotting tool that allows for plotting of FlowNet ensembles and observations.

Fixes
- [365](https://github.com/equinor/flownet/pull/365) Fixes bug where the parquet parameters file would not be saved for iterations larger than 9.

Changes
- [365](https://github.com/equinor/flownet/pull/365) Structure change of the code. Moved all forward models called from ERT to a seperate folder ert/forward_models. Scripts moved: delete_simulation_output, save_iteration_parameters, iteration_analytics, render_realization and flow_job.
- [347](https://github.com/equinor/flownet/pull/347) Additional flow nodes is now allowed to be either a list (equal length of number of layers) or a single integer (which will be split over the layers according to volume of concave hull).

Fixes
- [346](https://github.com/equinor/flownet/pull/346) Fixes bug where the analytics module would fail for iterations larger than 9 (i.e., iteration number with two or more digits).

Added
- [341](https://github.com/equinor/flownet/pull/341) Added option to add a min permeability threshold for which tubes are deactivated when the permeability is below this value.
- [305](https://github.com/equinor/flownet/pull/305) Functionality of generating seperate flownets per layer is enabled, additional nodes are split between the layers according to the volume of concave hull of the layers.
- [298](https://github.com/equinor/flownet/pull/298) Connections between (well)nodes that go through non-reservoir are now removed. Angle threshold export to user.
- [296](https://github.com/equinor/flownet/pull/296) Adapted perforation strategy to allow for layering + bug fixes in the 'multiple' and 'multiple_based_on_workovers' perforation strategies.
- [284](https://github.com/equinor/flownet/pull/284) Added the option to specify cumulative phase rates as observations (WOPT, WWPT, WGPT, WGIT, WWIT)

Fixes
- [339](https://github.com/equinor/flownet/pull/339) Fixes bug where the analytics module would fail when a well name filter was supplied.
- [325](https://github.com/equinor/flownet/pull/325) Fixes bug where the analytics module would repeat the 8th iteration analytics for iterations larger than 9.
- [256](https://github.com/equinor/flownet/pull/256) Fixes issues with duplicate names in hyperopt by using the full path in yaml for hyperopt parameter names.
- [272](https://github.com/equinor/flownet/pull/272) Adds resampling of observation dates at requested frequency by finding nearest date among existing observation dates (i.e., no interpolated dates added)

Changes
- [342](https://github.com/equinor/flownet/pull/342) Additional flow nodes are now a list in the configuration file, which allows for variation in the number of addtional nodes from layer to layer.
- [337](https://github.com/equinor/flownet/pull/337) You can now add multiple analytics workflows in your FlowNet config. This is also a breaking change in that it requires you to change the ert.analysis keyword to being a list.
- [322](https://github.com/equinor/flownet/pull/322) RSVD input through csv files can now be done either as one table used for all EQLNUM regions, or as one table for each EQLNUM region. The csv file needs a header with column names "depth", "rs" and "eqlnum" (the latter only when multiple tables are defined).

Added
- [251](https://github.com/equinor/flownet/pull/251) The number of available prior distributions have been expanded. It is now also possible to define the following probability distributions in the config yaml: _normal_, _lognormal_, _truncated normal_, _triangular_, _constant_, in addidion to the prevously defined ones _uniform_ and _loguniform_. This can be done for the following parameters: _bulk volume multiplier_, _porosity_, _permeability_, _aquifer size_, _fault multipliers_, _contacts (owc, goc, gwc)_ and _relative permeability parameters_.
- [234](https://github.com/equinor/flownet/pull/234) Added [mlflow](https://www.mlflow.org/) in combination with [hyperopt](https://github.com/hyperopt/hyperopt) which allows running flownet in batch to explore and optimise hyperparameters.
- [221](https://github.com/equinor/flownet/pull/221) For traceability and reproducibility, the FlowNet configuration file is now automatically copied to the output folder. The content of 'pip freeze' is stored in the file 'pipfreeze.output' in the output folder.
- [220](https://github.com/equinor/flownet/pull/220) User can now provide the historical _control mode_ in the production wells and _control mode_/_target_ in the injection wells. One value for each, applied to all wells.
- [199](https://github.com/equinor/flownet/pull/199) Added the possibility of defining a 'base' value for all parameters related to relative permeability. This provides the opportunity to interpolate between three relative permeability models. All the 'min' values will be used to construct a pessimistic or low model, the 'base' values are used for a 'base' model, and the 'high' values are used for an optimistic or high model value. The history matching can then be done with only one (two for three phase models - oil/water and gas/oil interpolated independently) relative permeability parameter(s) per SATNUM region.
- [197](https://github.com/equinor/flownet/pull/197) Added opening/closing of connections based on straddling/plugging/perforations through time. This PR also adds a new perforation strategy `multiple_based_on_workovers` which models the well connections with as few connections as possible taking into account opening/closing of groups of connections.
- [188](https://github.com/equinor/flownet/pull/188) The possibility to extract regions from an input simulation model extended to also include SATNUM regions. For relative permeability, the scheme keyword can be set to 'regions_from_sim' in the configuration file.
- [189](https://github.com/equinor/flownet/pull/189) User can now provide both a _base_ configuration file, and an optional extra configuration file which will be used to override the base settings.
- [155](https://github.com/equinor/flownet/pull/155) Adds reading of simulation 'well logs' to condition the priors of permeability and porosity based using kriging

Changed
- [199](https://github.com/equinor/flownet/pull/199) Removed deprecated parameters in pyscal ('krowend', 'krogend') from config file. Added 'kroend' to config file.
- [228](https://github.com/equinor/flownet/pull/228) FlowNet is now `pip` installable without any dependency compilations (also the custom ERT forward model running Flow is installed automatically). Note that any `simulator` part of configuration files need to be removed.

Added
- [160](https://github.com/equinor/flownet/pull/160) Adds the possibility to extract regions from an existing model when the data source is a simulation model. For equil, the scheme key can be set to 'regions_from_sim'
- [157](https://github.com/equinor/flownet/pull/157) Adds a new 'time-weighted average open perforation location' perforation strategy called `time_avg_open_location`.
- [150](https://github.com/equinor/flownet/pull/150) Adds this changelog.
- [146](https://github.com/equinor/flownet/pull/146) Added about page to documentation with logo of industry and research institute partners.
- [138](https://github.com/equinor/flownet/pull/138) Print message to terminal when the schedule is being generated instead of utter silence.
- [134](https://github.com/equinor/flownet/pull/134) Egg [model](https://github.com/equinor/flownet-testdata/blob/master/egg/ci_config/assisted_history_matching.yml) added in CI.
- [128](https://github.com/equinor/flownet/pull/128) Add possibility to use the original simulation model volume as volumetric constraints for adding new nodes as well as the option to set a maximum connection length.
- [115](https://github.com/equinor/flownet/pull/115) Documentation of the preliminary Egg model results have been added.
- [104](https://github.com/equinor/flownet/pull/104) Adds a new FlowNet logo.
- [98](https://github.com/equinor/flownet/pull/98) Rock and Aquifer keywords are now allowed to be optional parameters.
- [95](https://github.com/equinor/flownet/pull/95), [#97](https://github.com/equinor/flownet/pull/97) Relative permeability and `training_set_end_date` are now allowed to be optional. The training set definition will be checked upon parsing the configfile.
- [90](https://github.com/equinor/flownet/pull/90) Running the analysis workflow is now optional.
- [77](https://github.com/equinor/flownet/pull/77) Adds a warning message when the `WSTAT` keyword is not present in the original simulation data deck. The default status is `OPEN`.

Fixed
- [141](https://github.com/equinor/flownet/pull/141) Fixed misplacement of `WCONINJH` target (`RATE` / `BHP`). The default `RATE` is however still always used.
- [137](https://github.com/equinor/flownet/pull/137) After having moved all example data to [flownet-testdata](https://github.com/equinor/flownet-testdata) the examples folder had no use longer. Removed.
- [136](https://github.com/equinor/flownet/pull/136) The ERT queue name was wrongly required in the config file. It is now allowed to have no value.
- [119](https://github.com/equinor/flownet/pull/119) Workaround for [equinor/libres#984](https://github.com/equinor/libres/issues/984)
- [92](https://github.com/equinor/flownet/pull/92) Fixed problem where the last production well would not correctly be shut.
- [88](https://github.com/equinor/flownet/pull/88), [#106](https://github.com/equinor/flownet/pull/92), [#108](https://github.com/equinor/flownet/pull/108), [#110](https://github.com/equinor/flownet/pull/110) Fixed bugs in parsing the config that were a result of a changed behavior in configsuite where an item that has subitems with subitems, no longer can be tested for all `None` as the subitem with subitems in fact is a `NamedDict` with all `None`.
- [89](https://github.com/equinor/flownet/pull/89), [#91](https://github.com/equinor/flownet/pull/91), [#99](https://github.com/equinor/flownet/pull/99) Updates and fixes in git workflow.
- [81](https://github.com/equinor/flownet/pull/81) Various bugfixes in the analysis workflow are now solved.
- [75](https://github.com/equinor/flownet/pull/75) Resampling via the config path is no longer supported. This PR removes also the resampling option from the config file parser.

Changed
- [147](https://github.com/equinor/flownet/pull/147) The observation uncertainty now only needs to be specified for the vectors that are actually used in the conditioning. This PR also adds possibility of defining uncertainty for WWIR and WGIR into the flownet config YAML.
- [132](https://github.com/equinor/flownet/pull/132) The observation uncertainty, used for conditioning models, is now exposed to the user via the configuration file instead of hidden in the `observations.yamlobs.jinja2` template. This PR also introduced a wider usage of passing around the `ConfigSuit.snapshot`.
- [128](https://github.com/equinor/flownet/pull/117) Simulation models and configuration files used in CI are now coming from [flownet-testdata](https://github.com/equinor/flownet-testdata) and are thus no longer an integral part of the flownet repository.
- [125](https://github.com/equinor/flownet/pull/125) AHM and prediction `subprocess` call are now being performed through a common `run_ert_subprocess` function call.
- [123](https://github.com/equinor/flownet/pull/123) Significant speed-up of the fault raytracing by limiting the ray tracing to the bounding box of individual connections instead of ray tracing over all triangles of all faults for each connection.
- [111](https://github.com/equinor/flownet/pull/111) Phases present in the FlowNet model are to be set manually in configuration by the user.
- [63](https://github.com/equinor/flownet/pull/63) The default simulator name has now been changed to _Flow_.
- [80](https://github.com/equinor/flownet/pull/80) FlowNet now supports relative paths in the config file.