MODULE 3. MODELS

This section provides a centralized view of all created fit cases and supports the creation of new models within the platform.

Figure 10. Models Section.

Top bar widgets:

Figure 11. Model Top Buttons.

Categorization

Enables users to filter or group models based on the hierarchy defined in the environment (e.g., by field or asset).

**Refresh
**Reloads the current view or undoes recent changes.

Customizing Columns

Users can personalize and arrange the display columns to suit their workflow preferences. Available columns include:

• Name. Name attributed to the model.
• Description. Description of the model.
• Asset. The hierarchical entity (e.g., field) associated with the model.
• Dataset. Identifies the dataset used during model creation.
• Status. Displays the current model state:
  • Running.
  • Pending.
  • Failed.
  • Terminated.
• Fit Type. Specifies whether the model is a Back Test (BT) or Full Fit (FF).
• Model Type. Describes the type of enhanced oil recovery (EOR) used: Waterflood, Gas Injection, Polymer flood or Water Disposal.
• Created Date. The date the model was generated.
• End Date. The completion date of the model run.
• Run Time. The total time required for model execution.
• Model Version. Indicates the LOCSIM version.
• Parent. Indicates the parent entity in the hierarchy.
• Flow Unit Type: Indicates whether the model includes no separation or separation into compartments/fault blocks.
• Multi Reservoir: Specifies if the model includes multiple reservoirs.
• Multiplier: Indicates if well-level multipliers have been applied.
• Optimization: Indicates the optimization case created from a model.

Compare Models

The Model Comparison feature lets users view two models side by side. Select exactly two models using the checkboxes to enable the Compare option and analyze their results simultaneously.

Bulk Actions

Perform multiple actions for a group of models selected. Options are:

• Mark as Favorite.
• Marked as Locked.
• Marked as Restricted.
• Marked as Target.

Search Bar

Allows users to enter keywords or phrases to quickly locate relevant models.

New Model

Use the New Model button to launch the model creation wizard and configure a new fit case.

Chart columns and options:

Checkbox

Enables the selection of one or multiple models for batch actions.

Column Names

The table columns reflect the attributes listed in the Customizing Columns section, helping users visualize model data at a glance.

Three-dot button:

The three-dot button beside each model provides quick access to advanced actions, offering a clean and intuitive interface. Depending on the model's status and type, the following options may be available:

• Edit
  • Modify Name & Description.
• Data
  • Sync Post Processing.
• Workflows
  • Run Full Fit.
  • Scenarios / Run Scenarios
  • Run Optimization.
• Actions
  • Toggle Favorite.
  • Toggle Restricted.
  • Toggle Target.
  • Copy Fit.
  • Assign Asset.
• Remove Model.

Available options may vary based on the model's type and current status.

3.1 NEW MODEL CREATION

The Model Wizard guides users through a 10-step process to configure and run a new fit case. This streamlined workflow is designed to modernize and enhance traditional simulation methodologies.

3.1.1 General

3.1.1.1 What type of Model do you want to train?

Figure 12. General Tab: Type of Model.

In this tab is set the model that is desired to be trained, which can be: Waterflood,Miscible Gas Injection, Polymer Injection, Salt Water Disposal, Primary Production, CO2 Injection, Water Alternating Gas (WAG) or Proxy Modeling.

3.1.1.2 Now, give your model a name and a brief description

Figure 13. General Tab: Name and Description.

Characterization of the model, the options displayed are:

• Name. Case name.
• Description. Optional text to describe the model.
• Use the latest version of Locsim. Automatically selects the latest LOCSIM version.
• Show Non-Production Contracts. Enable the visualization of special branches.
• Locsim. Display the list of available versions and branches (branches shown only if the Non-Production Contracts toggle is activated).

3.1.2 Dataset

Figure 14. Dataset Tab.

Upload or choose an existing dataset

Enables to upload a new dataset or select a preexisting dataset.

3.1.2.1 Upload a new dataset

If selected to upload a new dataset a button is facilitated to emit this action.

Figure 15. Upload New Dataset.

Upload Button

Enables the user to upload the new dataset.

Name

Dataset Name.

Description

Optional text to describe the dataset.

Field

Indicates the field the datasets belong to.

Data Files

Upload ‘Completion’ and ‘Production’ files individually. (Refer to the Appendix for format requirements).

Choose an existing dataset button (Review the section 3.1.2.2 Use an existing dataset)

Show Optional Files button

Option that enables uploading additional files such as: Histogram, VFP tables, Petrophysical Properties Data and/or Fault file.

Once the dataset has been selected and/or uploaded, the ‘Next’ button enables the user to continue configuring the case.

3.1.2.2 Use an existing dataset

Search bar

Allows users to enter keywords or phrases to quickly locate relevant datasets.

Datasets list

In case previous datasets have been uploaded, they will be displayed on a list, which will allow the selection of any.

Upload a new dataset button (Review the section 3.1.2.1 Upload a new dataset)

Show Optional Files button

Option that enables uploading additional files such as: Histogram, VFP tables, Petrophysical Properties Data and/or Fault file.

Once the dataset has been selected and/or uploaded, the ‘Next’ button enables the user to continue configuring the case.

3.1.3 Wells

This tab visualizes all wells using data from the uploaded dataset.

Map

The generation of a geographical map illustrates the spatial distribution of wells based on the coordinates specified in the completion file for each individual well.

Figure 16. Wells Tab.

Map Type

The geographic map is displayed per well type, fault block or compartment.

Reservoir

Enables the wells geographical display per reservoir (option only available for multi reservoir models).

Fault Blocks

Enables the wells geographical display per fault blocks (option only available for multi reservoir models).

Selected Wells

Allows to select or deselect a certain region/group of wells that are not going to be considered in the analysis.

3.1.4 Production

Figure 17. Production Tab.

Visualizes historical production data across multiple phases with flexible graph configurations.

Left side panel:

Figure 18. Left Panel Configuration.

• Chart Type. Charts can be compared or combined.
  • Combine.
  • Compare.
• Top Measurement. Enables the measurement selection that will be displayed in the top graph.
  • GLR.
  • GOR.
  • Gas.
  • Gas Injection.
  • Gas Pressure.
  • Liquid.
  • Oil.
  • Polymer Concentration.
  • Pressure.
  • WOR.
  • Water.
  • Water Injection.
• Bottom Measurement. Enables the measurement selection that will be displayed in the bottom graph.
  • GLR.
  • GOR.
  • Gas.
  • Gas Injection.
  • Gas Pressure.
  • Liquid.
  • Oil.
  • Polymer Concentration.
  • Pressure.
  • WOR.
  • Water.
  • Water Injection.
• Unit of measurement. Field or Metric units.
• Plot type. Rate or cumulative vs time.
• Select Wells. Permit select or unselect a group of wells.

Figure 19. Select Wells.

Search bar

The search bar allows you to quickly find the relevant wells within the list.

Select All Wells

The Wells table enables the selection of all the wells at the same time by using the checkbox next to the wells name.

Reservoir button

Wells associated with a certain reservoir (in the case of the reservoir model) are selected automatically through this option.

Wells List Table

Selection of all or a group of wells (injectors or injectors and Producers) desire to be targeted during the scenario evaluation.

Your Selection Table

Displays the chosen well selection that will be taken into account during the scenario evaluation.

• Plot Wells as. Selection of how the plots should be visualized, rather:
  • Combined.
  • Separated.

Graphs:

Figure 20. Production Graphs.

Visual representation of the historical data of the measurement selected, showcasing its performance throughout time. Components include:

• Leftt Y-axis. Measurement selection.
• X-axis: Time.
• Right Y-axis. Well count.
• Three dots. Enables the following options:
  • View in full screen.
  • Print chart.
  • Download PNG Image.
  • Download JPEG Image.
  • Download PDF Document.
  • Download SVG Vector Image.
  • Download CSV.
  • Download XLS.
  • View data table.

3.1.5 Fit Type

What type of fitting do you want to use for your model?

The waterflood project incorporates a train and test algorithm during the fitting phase, implementing two types of fitting:

Figure 21. Fit Type Segment.

Back Test

The algorithm utilizes 50-70% of the historical data uploaded during the training period to adjust fitting parameters (both global and local). The remaining portion of the data is then used in the testing period, validating the model's predictivity. Once the user is satisfied with this adjustment, they proceed to execute the second fitting.

Full Fit

In this fitting, the algorithm employs 90%-99% of the historical data uploaded in the training period, focusing on adjusting dynamic parameters (pressure and responsivity). The remaining 1%-10% of the data is utilized in the testing period to validate the model's predictivity. Important Note: It is recommended to run a Back Test first, followed by a Full Fit, ensuring the fit model is executed in the recommended workflow. The subsequent module provides a detailed explanation of the initial steps involved in uploading, creating, and running a case.

3.1.6 Setup

Figure 22. Setup Segment.

Multi Reservoir Model

True if it is a multi-reservoir fit False if is a well-level fit.

Training Start Date

Date for the case to start the training.

Training End Date

Date for the case to end the training.

Backtest End Date

Date to end the Back Test from the start date. It is recommended to select a number of months representing 20-30% of the total number of months with data included in the training.

Backcast Start Date

Backcast evaluation start date designation.

Model Horizontal Wells

If true, load completion lengths, where each point in space for the horizontal wells will be used separately for rate calculations, improving the horizontal well modeling. If set to false, the locations will be the first point of the well.

Injection Water Loss Percentage

Water Injection loss factor varies from 0 to 1, 0 = No loss, 1 = All injection is lost.

Injection Gas Loss Percentage

Gas Injection loss factor varies from 0 to 1, 0 = No loss, 1 = All injection is lost.

Maximum Gas Lift Pressure (Delta) for dBHP vs dGLR model

Maximum gas lift pressure calculated from GLR. Igored by using the default value of 0.

Boundary Type

Options are:

• Closed boundary.
• Edge Aquifer.
• Bottom Aquifer.

Producer Pressure Control Mode

Producer well control mode, see options below.

• BHP control for all producers.
• BHP from dBHP vs dGLR model with universal constant.
• BHP from dBHP vs dGLR model with well by well constant.
• BHP from dBHP vs dGLR model using historical data with universal constant.
• BHP from dBHP vs dGLR model using historical data with well by well constant.
• BHP from dBHP vs dGLR model, except wells with conduits which use dBHP vs dInjRate model.

Producer Control Mode

Producer well rate control mode, see options below.

• BHP or VFP Control
• Oil Rate Control
• Liquid Rate Control

Reserves Data Source Type

When running a multi-reservoir model (multi-layered), refinement of the initialization of some of the parameters can be done per reservoir. Ranges for water and gas initial saturations, initial pressures, and NetPay can be included. Options enabled are:

• Use constant net pay. Use the same net pay set in the configurations tab for all the wells.
• Use reserves data table for initial saturations and net pay per layer. Applies the data uploaded in the Reserves Data table configuration (must be complete even for single layer level analysis).
• Use reserves data table for initial saturations per layer and net pay per well from completions file. Applies the data uploaded in the Reserves Data configuration (must be complete even for well level analysis), taking from it the saturation and pressure ranges setup and the net pay from the completions input.

Boundary Distance

Distance to the boundary from the convex hull of wells. There are a maximum Boundary Distance = 3 times the grid distance (dS).

Flow Unit Type

Option to enable separating the field into different flow units. No separation=0, Separate into Compartments=1, Separate into fault blocks=2, Custom=3. For fault blocks or compartments, data is provided in the completion input file.

Gravity Effect

Enable/disable gravity effects. Consider this option if the dip angle of the reservoirs are considerable.

Initialization Mode

System state initialization mode. It can be set as:

• Constant initialization.
• Gravity based initialization. Using this option, each layer’s pressure is initialized using the reference depth of each layer.

Historical Graphs

The charts on the right enable a visual representation of the historical performance of the different phases (oil, water, gas, water injection, pressure) and pressure trends over time. After the training dates along with the Back Test end date have been selected, the data selection will be reflected in these charts.

3.1.7 Constants

Define PVT, Rock Fluid, and Reference properties that are treated as constants in the following window. Although parameters are defined as constant values, these will vary with pressure and saturation variations as defined within the underlying system state maps.

Figure 23. Constants Segment.

3.1.7.1 PVT Properties

Oil Compressibility

Oil Compressibility [1/psi].

Reference Water Viscosity

Water reference viscosity [cP].

Min Water Viscosity

Water minimum viscosity [cP].

Water Compressibility

Water Compressibility [1/psi].

Oil Density

Oil Density [lb/ft^3].

Water Density

Water Density [lb/ft^3].

Bubble Point Pressure

Bubble point pressure [psi].

Solution Gas Ratio at Pb

Solution Gas [mscf/bbl].

Formation Volume Factor at Pb

Formation volume factor at bubble point [resbbl/surbbl].

Min Miscibility Pressure

Minimum Miscibility Pressure (MMP) [psi].

3.1.7.2 Rock Fluid Properties

Corey Exponent for Gas

Gas Relative Permeability model exponent.

_
Corey Exponent for Oil_

Oil Relative Permeability model exponent.

Corey Exponent for Water

Water Relative Permeability model exponent.

_
Critical Gas Saturation_

Residual Gas Saturation [fraction].

_
Residual Oil Saturation_

Residual Oil Saturation after water flooding [fraction].

_
Miscible Gas Injection Sor_

Residual Oil Saturation after gas flooding [fraction].

SWD K from Phi (A)

K = A * exp(10,N*Phi)

SWD K from Phi (N)

K = A * exp(10,N*Phi)

Stone 1 KrOG End

Stone 1 Gas-Oil End Point.

Stone 1 nOG Exponent

Stone 1 Gas-Oil curve Exponentil.

Kv/Kh

Kv/Kh between layers and reservoirs.

Connate Water Saturation

Connate Water Saturation [fraction].

3.1.7.3 Reference Properties

Universal Gas Lift Model Coefficient

Coefficient of universal gas lift model.

Reference Temperature

Reference temperature [°F].

_
Reference Pressure_

Reference pressure [psi].

_
Net Pay_

Thickness of reservoir [ft].

_
Water Oil Contact_

Water oil contact. 0 = no water leg [ft].

_
Gas Oil Contact_

Gas oil contact. 0 = no gas cap [ft].

_
Reference Depth for Initialization_

Reference depth for PoInit for initialization. 0 = use average completion depth.

_
Aquifer Pressure_

Aquifer pressure [psi].

_
Reservoir Initial Salinity_

Reservoir Initial Salinity [ppm].

Reservoir Dip Angle

Dip angle of reservoir top [°].

_
Reservoir Dip Azimuth_

Azimuth angle of reservoir top [°].

_
Reference X Location of Top_

Reference X of the default reservoir top [ft].

_
Reference Y Location of Top_

Reference Y of the default reservoir top [ft].

_
Reference Z Location of Top_

Reference Y of the default reservoir top [ft].

3.1.8 Uncertainties

Configure PVT, Rock Fluid, and Initialization parameters that will be used within ranges. As the model is trained, and the underlying system stage maps are adjusted, the parameters listed in this list will be adjusted accordingly and automatically through EnKF (ensemble Kalman filter).

Figure 24. Uncertainties Segment.

3.1.8.1 Initialization Parameters

Initial Pressure

Initial reservoir pressure at time of fit start. The model is initialized randomly with pressure within the provided range [psi].

Initial Water Saturation

Initial reservoir water saturation at the time of fit start. The model is initialized randomly with water saturation within the provided range [fraction].

Initial Gas Saturation

Initial reservoir gas saturation at the time of fit start. The model is initialized randomly with gas saturation within the provided range [fraction].

3.1.8.2 Rock Fluid Properties

Porosity

Reservoir porosity range [fraction].

Permeability

Reservoir permeability range [mD].

Rock Compressibility

Compressibility of rock (change in rock volume as a function of pressure), the larger the number, the more compressible the rock [1/psi].

Oil Rel Perm End Point

Endpoint relative permeability of water, the relative permeability at max oil saturation [fraction].

Gas Rel Perm End Point

Endpoint relative permeability of gas: the relative permeability at max gas saturation [fraction].

Water Rel Perm End Point

Endpoint relative permeability of gas: the relative permeability at max water saturation [fraction].

Aquifer Vertical Permeability

Aquifer vertical permeability [mD].

3.1.8.3 PVT Properties

Reference Oil Viscosity

Oil viscosity at reference pressure [cP].

Min Oil Viscosity

Minimum oil viscosity, which is used in the viscosity model to ensure an always positive viscosity [cP].

Reference Gas Viscosity

Gas viscosity at a reference pressure [cP].

Min Gas Viscosity

Minimum gas viscosity, which is used in the viscosity model to ensure an always positive viscosity [cP].

Max Miscibility

Maximum miscibility (Omega coefficient) reached at Minimum Miscibility Pressure (MMP [psi]).

Zero Miscibility Pressure

The lower end of reservoir pressure below which, gas has zero miscibility with oil (Omega = 0) [psi].

3.1.9 Solver

Figure 25. Solver Segment.

3.1.9.1 Solver

Configure solver settings that define how the engine operates in solving the equations. Define as well the data uncertainty parameters (standard deviation) and bounds for the underlying system state maps.

Fit Level

What type of aggregated data to use for fitting. Clustering helps in improving overall fit quality by reducing the effect of noise and allowing a smaller ensemble to be effective. The fitting is done well by well regardless of fitType. Options are:

• Field Level (Oil, Gross, Gas, Water Injection).
• Each Well (Oil, Gross, Gas, Water Injection).
• Each Cluster (Oil, Gross, Gas, Water Injection).
• Each Cluster (Oil, Gross, Gas, Water Injection, Reservoir Pressure).

The preferred option is to select ‘Each Cluster: Oil, Gross, Gas and Water Injection’.

_
Geomodel EnKF Type_

How EnKF fit concept uses when multiple geomodels present in the production data.

• Mixed geomodel update, fit average geo-data.
• Mixed geomodel update, fit indivdual geo-data.
• Separate geomodel update, fit indivdual geo-data.

Fitting Phase

0 = Fit oil rate, 1 = fit gross rate, 2 = fit gross and gas rate.

_
Number of Clusters_

Number of clusters used to aggregate data to fit. Wells are clustered on distance-based (considering FaultBlocks separation). As a rule of thumb, the number of clusters is set to Number of wells/10. Maximum number of clusters is recommended around 20.

_
Number of Ensembles_

Number of ensemble members of the Ensemble Kalman Filter; the larger the number, the better the uncertainty quantification, but is also more computationally intensive. Recommended and by default is 96 members.

_
Neural Network Weights_

Percent of wells to be used as the number of parameters for reservoir responsivity. These are the parameters of the neural network being trained to create the reservoir responsivity map: the larger the number, the better the model will fit to well-level data, but this can also lead to poorer predictions due to overfitting. Recommended to set 95.

_
Phase Fitting Weight_

Scaling factor for weighting oil rate vs water injection rate in reservoir responsivity calculation. 1= equal weight, 0 = fully oil weighted. Helps in improving oil rate fit as injection rates can be much larger than production rates. Beta is total injection / total liquid production. It is the weight to adjust responsivities to match Liquid production and Injection rates. As we use the same responsivity map for both matching injection rates and production rates, we need to establish which phase (injection rates or liquid rates) to give more weight to when modifying the responsivity. Because the injection is exactly matched most of the time, the injection influence is almost zero regardless of beta. This parameter will be meaningless unless the injection is not being matched.

3.1.9.2 Standard Deviations

Measurement Error

sd: Oil Rate

Oil Rate Uncertainty.

_
sd: Gross Rate_

Gross Rate Uncertainty.

sd: Gas Rate

Gas Rate Uncertainty.

_
sd: Water Injection Rate_

Water Injection Rate Uncertainty.

_
sd: Reservoir Pressure_

Reservoir Pressure Uncertainty.

3.1.9.3 State Bounds

**_Max Pressure
_**Maximum allowed model pressure [psi].

Min Pressure

Minimum allowed model pressure [psi].

**_Max Water Saturation
_**Maximum allowed water saturation [fraction].

**_Min Water Saturation
_**Minimum allowed water saturation [fraction].

**_Max Gas Saturation
_**Maximum allowed gas saturation [fraction].

**_Min Gas Saturation
_**Minimum allowed gas saturation [fraction].

3.1.10 Data Edits

Figure 26. Data Edits Segment.

3.1.10.1 Data Edits

Use the Data configuration options in this section to avoid missing values and outliers

Compress Fit Data (ESMDA)

Compress the fitting data for ESMDA.

Interpolate Injection

Whether to interpolate injection data when data is missing. If this is true, linear interpolation is used to interpolate missing data: otherwise, the missing data is set to 0.

Interpolate Production

Whether to interpolate production data when data is missing. If this is true, linear interpolation is used to interpolate missing data: otherwise, the missing data is set to 0.

Filter Outliers

Whether to apply median filtering to data: this is used to remove outliers in data.

Extrapolation Timesteps

Maximum number of days to forward extrapolate missing data from the end of DB data.

Long Interpolation to Remove

Maximum number of days to remove for long interpolations.

Override Water Injector BHP

If greater than minimum allowed state pressure and less than maximum allowed state pressure, will override BHP from DB for injectors.

_
Override Gas Injector BHP_

If greater than minimum allowed state pressure and less than maximum allowed state pressure, will override gas injection BHP from DB for injectors.

_
Override Producer BHP_

If greater than minimum allowed state pressure and less than maximum allowed state pressure, will override BHP from DB for producers.

_
Override Min Producer BHP_

If greater than minimum allowed state pressure and less than maximum allowed state pressure, will override minBHP from DB for producers [psi].

Override Water Injector Max BHP

If greater than minimum allowed state pressure and less than maximum allowed state pressure, will override BHP from DB for water injectors.

_
Override Gas Injector Max BHP_

If greater than the minimum allowed state pressure and less than the maximum allowed state pressure, will override BHP from DB for gas injectors.

Override Injector Responsivity

If greater than zero, responsivity for injections will be overridden with the input value. If zero, responsivity at injectors will be calculated from the data.

VFP BHP Damping

Damping factor for VFP BHP generated in the previous time steps.

Min Producer BHP

Minimum producing BHP as input in constraining the model. The purpose is to remove outliers in the data (from any sources). The default is 0 when the limit is not used.

_
Max Producer BHP_

Maximum producing BHP as input in constraining the model. The purpose is to remove outliers in the data (from any sources). The default is 0 when the limit is not used.

_
Min Injector BHP_

Minimum injecting BHP as input in constraining the model. The purpose is to remove outliers in the data (from any sources). The default is 0 when the limit is not used.

Max Injector BHP

Maximum injecting BHP as input in constraining the model. The purpose is to remove outliers in the data (from any sources). The default is 0 when the limit is not used.

Min Producer Completion Length

Minimum perforation length for producers. If the producer has a completion length below this number, then field level average will be set. Ignored by using the default value of 0.

Min Injector Completion Length

Minimum perforation length for injectors. If the injector has a completion length below this number, then field level average will be set. Ignored by using the default value of 0.

Max Producer Completion Length

Maximum perforation length for producers. Ignored by using the default value of 0.

Max Injector Completion Length

Maximum perforation length for injectors. Ignored by using the default value of 0.

3.1.11 Review

This segment displays all the information submitted of the case parameters and properties setup. At the bottom of the tab, the Create Model button will allow running the model after reviewing the configuration established.

Figure 27. Review Segment.

Advanced options are included in this tab. The options included in the Advanced section are described below:

3.1.11.1 General Setup

Well Multipliers

Multipliers can be used to refine the fit for individual wells. The multipliers increase or decrease proportionally the productivity or injectivity of a well. The default value is 1 when not specified. To include multipliers, click on the pencil symbol, and automatically a window will pop up showing the list of the well (producers and injectors), bore, reservoir, and a multipliers column, this last header contains a blank box per well so the user can add the multipliers that consider necessary, those that do not require will be left blank. The Apply button enables the changes made or Clear to be undone. For these changes to take place, multiplierMode must be set to True.

Layer Data

Option enabled when activating through the reserves Data Source parameter (Setup Tab) the use reserves data table options. Allows the inclusion of parameters per reservoir.

Figure 27. Initialized Reserve Data.

• Add Reserves. Allows including reservoir level parameters, enabling a window chart that allows provisioning of the So Init Min, So Init Max, Sg Init Min, Sg Init Max, Net Pay, Po Init Min, Po Init Max, Porosity Mult, Permeability Mult, KrW Mult, KrO Mult, KrG Mult, Swc, Sor, Sgc, nW, nG, Water Injection Loss, and Gas Injection Loss per layer.
• Cancel. Cancel the modifications made.
• Save. Save the modifications made.

3.1.11.2 Setup

Machine Size for Computation
Choose size of the machine to run fit. The options are:

• Auto.
• Small.
• Large.

Allow Conversions

If true, include in fit historical production/injection data from converted wells.

Forecast Calibration Phase

Target phase for production correction with BHP. Options are: 0 = Oil, 1 = Liquid. This correction mode is utilized at the end of the FullFit to calibrate the start point of the forecast to the last available data; the selected phase will be the one utilized for the calibration.

Forecast Calibration Mode

0 = No Correction, 1 = Responsivity Only, 2 = BHP Only, 3 = BHP and Responsivity. The selection of the Calibration Mode will be utilized for correcting the initial production rate (oil or liquid) with the last available historical data. This correction mode is utilized at the end of the FullFit to calibrate the start point of the forecast to the last available data.

_
Forecast Calibration Iterations_

Number of iterations for production correction.

**_Forecast Calibration Time Steps
_**Number of time steps of last historical production data to be used for production correction

Forecast Calibration Use Full Ensemble

If true full ensemble will be used to calibrate production, otherwise the P10-P50-P90 will be used

3.1.11.3 Conduits

Apply Conduits

Enables the application of the conduit evaluator in the model.

Conduit Gas Direct Fac

Well API Identifier

maximum Responsivity

minimum Responsivity

Stop State Update Date

Conduits Mode

0 = Do not apply, 1 = Apply as listed in conduits physically, 2. Apply conduits with direct gas cycling.

Conduit List

Load the conduits into the model using the chart provided.

Add Conduit button

Activates row-by-row loading of the desired conduits into the model for better adjustment.

Conduits Chart

The chart is constituted by the following columns:

• InjectorAPIs
• ProducerAPIs
• width
• CondK
• Matrix Permeability
• Conduit Producer Completion Fraction
• Conduit Injector Completion Fraction

Cancel button

Discards any unsaved changes and closes the current window or dialog.

Save button

Stores the current changes and updates the configuration or document.

3.1.11.4 Solver

EnKF Inflation Factor

Inflation factor for Ensemble Kalman Filter: increasing the inflation factor helps in preventing ensemble collapse, which is the tendency of all the ensemble members to become similar over time.

EnKF Ensemble Perturbation

Whether to perturb ensemble members after Kalman Update, this acts like traditional inflation. Another option to add noise to model input data and prevent artificial collapse of ensemble. If data is of suspect quality, it is recommended to check this option as true.

Responsivity Smoothing Weight

Magnitude of the weight of second order smoothing of reservoir responsivity map, the larger the weight, the smoother the map. Smoothing parameter of the reservoir responsivity map (essentially the characterization of reservoir properties). Larger values correspond to more smoothing, i.e. less heterogeneity/property variation. Values too low or too large can result in poor predictions for new wells.

Responsivity Smoothing Type

0 = Normal smoothing, 1 = Least square smoothing, 2 = SVD smoothing, 3 = Compressed Smoothing.

HW Responsivity Mode

The options are the following:

• Normal smoothing
• Least square smoothing
• SVD smoothing
• Compressed Smoothing

Netpay Mode

The options are the following:

• NetPay not from property map. Not to override NetPay from completion csv with the map.
• Load From Property Map. Use NetPay from map.

Porosity Mode

The options are the following:

• Homogeneous. All cells have the same value in a certain model in the ensemble. Use the default porosity assignment provided by the application, where each EnKF ensemble realization assigns a unique value selected from the defined global range.
• Spatial Krigging. Use spatial variation with a krigging distance.
• Well Pilot Point. Use local values at pilot point method at wells with data as model parameters (SWD only for now).
• Load From Property Map. Use properties in the map.

Permeability Mode

The options are the following:

• Homogeneous. All cells have the same value in a certain model in the ensemble.
• Spatial Krigging. Use spatial variation with a krigging distance.
• Well Pilot Point. Use local values at pilot point method at wells with data as model parameters (SWD only for now).
• Load From Property Map. Use properties in the map.

Property Krigging Variogram Distance

Gaussian variogram distance for Krigging porosity and permeability fields, 0 = white noise.

Min Responsivity

Lower bound of the estimated responsivity. Used to prevent low injectivity and productivity.

_
Max Responsivity_

Upper bound of the estimated responsivity.

_
Min NetPay_

Lower bound of the estimated NetPay.

Max NetPay

Upper bound of the estimated NetPay.

Min Responsivity Multiplier for Production Calibration

Multiplier for lower bound of responsivity for forecast production calibration. This range is used if correction mode is utilized at the end of the FullFit to calibrate the start point of the forecast to the last available data; and if the Calibration Mode has been selected to be Responsivity.

Max Responsivity Multiplier for Production Calibration
Multiplier for upper bound of responsivity for forecast production calibration. This range is used if correction mode is utilized at the end of the FullFit to calibrate the start point of the forecast to the last available data; and if the Calibration Mode has been selected to be Responsivity.

Injector BHP Control Timesteps

The number of initial steps to run EnKF on BHP control, even if the specified control mode is rate control. This is in improving model quality.

Back Test Correction Factor Timesteps

Number of time steps at the end of the Back Test period used to calculate correction factors to debias predictions.

Full Fit Correction Factor Timesteps

The number of time steps at the end of Full Fit period used to calculate correction factors to debias predictions, if not provided uses nTCorr. During FullFit, 1 to 6 is recommended.

_
Disable EnKF State Update Timesteps (Full Fit Only)_

Number of time steps at the end of full fit for which the pressures and saturations will not be updated by EnKF.

Blind Test Timesteps

Number of time steps at the end of Full Fit for a blind test.

_
Exclude Empty Data_

Exclude missing data from enkf fit if true.

Ensemble Parallelization

Whether to run the Ensemble Kalman Filter in serial or parallel (recommended when having multilayer models).

Responsivity Parallelization

Whether to run Responsivity in serial or parallel (recommended when having multilayer models).

Observed Data Perturbation Level

Oberved Data Perturbation Level from zero to one. Choose 0 if would like accurate fit with less error.

EnKF Cluster with Flow Units

Create cluster and EnKF by domain.

EnKF GOR Update

Update Solution GOR at enkf analysis.

EnKF Rescale Saturation

Rescale saturations after enkf analysis.

Linear Interpolation of Responsivity

Use linear interpolation instead of RBF interpolation for responsivity at new wells.

EnKF Refine Saturation

Update saturations before responsivity.

BHP Moving Average

Maximum number of timesteps to apply moving mean to BHP data.

**_Timestep Size for Fit
_**Time step size in days.

Timestep Size for Prediction
Time step size in days for running prediction. Default value of 0 means fit timestep will be used.

Injectors BHP Control

Injector bottom hole pressure control: if true, injectors are run under bottom hole pressure control instead of rate control.

Deviated Well Index Factor

Deviated well factor in atan function, recommend 1. =0. without transform; > 0 compL = 1.56*fac*netpay * atan[compL/(1.56*fac*netpay)].

Min Distance to Merge Cells

Distance between wells below which the surrounding cells will be merged. When wells are closer to each other than the average distance between wells in the field, it is recommended to set this parameter to the average distance. Wells within a distance smaller than the one specified in this parameter will be merged into the same cell.

_
Length of Conduit Cells_

Distance of conduit cells. (use 0: use distance maped on matrix cells).

Oil Viscosity Multiplier

If true, use the oil viscosity multiplier from the completed csv.

HW Injection Distribution

If true, evenly distribute injection along HW; if false, use pressure distribution.

Apply Permeability Multiplier

If true, use a permeability multiplier from the completion file.

Responsivity Level

0 = Each flow unit has its own responsivity map, 1 = same map for each reservoir. The first option is recommended when multiple domains are utilized (multi-reservoir or different Fault Blocks).

Correction Factor Mode

0. No Correction; 1. Apply correction externally to prediction; 2. Apply correction internally to prediction; 3. Apply correction internally to prediction except for gas. This correction factor mode is applied at the end of the Fullfit process for field-level calibration to address possible prediction biases.

_
Backcast Mode_

Backcast with options of property and responsivity. The options are:

• Step Local Property and Responsivity
• Step Local Responsivity and Final Property.

Average Grid Distance

Approximate distance between wells: used to set the distance between internal points.

Neural Network Lambda

Neural network accuracy parameter: the larger the parameter, the more accurate the radius basis function kernels. The kernel is less well-conditioned, so it requires some trial and error to identify a good value.

_
Min Newton Iterations_

Minimum iterations of nonlinear Newton solver.

_
Max Newton Iterations_

Maximum iterations of nonlinear Newton solver.

_
Linear Solver Tolerance_

Linear solver tolerance: this is the accuracy to which the linear system of equations created by the nonlinear Newton solver is solved.

_
Convergance Tolerence of Pressure_

Convergence of boundary condition equations essentially determines how well will the control variables (injection rates or BHPs) be honored.

_
Convergance Tolerence of Gas_

Convergence of gas continuity equations, essentially the mass balance error tolerance for gas.

Convergance Tolerence of Oil

Convergence of oil continuity equations: essentially the mass balance error tolerance for oil component.

Convergance Tolerence of Water

Convergence of water continuity equations: essentially the mass balance error tolerance for water component.

Time Out

Timeout to kill model run: run will return as failed if runtime exceeds this time limit.

Drainage Distance

Locsim removes cells that are at a greater distance than the Drainage Distance from the closest producer.

Reservoir Sequence

Reservoir sequence in format A<B<C;D<E<F for each flow unit if reservoirs are vertically connected, else leave blank. Kv/Kh should also be defined for this feature to be active.

Interior Point Clearance Distance

Interior points closer than this distance from well points will be removed: which helps in improving the conditioning of the kernels. This parameter is used to increase the resolution of property maps, additional points added to the parameterization. If this value is too small, the model can become ill conditioned and lead to computational errors. If too big. Internal points and/or wells may be removed.

_
Remove Inactive Wells_

If true, remove inactive wells from the model.

Export Voronoi Plots

If true, export diagnostic Voronoi maps to results.

EnKF Localization Distance

Minimum distance for Kernel: used for temperature localization of the Ensemble Kalman Filter.

Max Condition Number

Maximum condition number of neural network kernel: the kernel is modified if the condition number is above this, such that the system can be solved accurately without convergence issues.

_
Dry Run Time Steps_

Number of time steps for dry run. Used to check if the model is initialized correctly with a short run.

Producer Well Model Distance

Well model distance parameter for producers: used with well model to calculate production rates from pressure distribution near the well.

_
Injector Well Model Distance_

Well model distance parameter for injectors: used with well model to calculate production rates from pressure distribution near the well.

_
Producer Well Model Decline Mode_

Various decline modes of gsFac, helps in better controlling decline rate of wells.

• Linear.
• Hyperbolic.

Producer Well Model Decline Exponent

Decline exponent for hyperbolic decline, 0: exponential, 1: harmonic.

Producer Well Model Decline Type

Producer well model options. Apply alphaDP to dpLift term only=3, Apply alphaDP to full pressure gradient= 4.

Number of Internal Points

Number of internal model points: these are in addition to the well and/or completion points. This helps in having more evenly distributed points throughout the model and improving kernel conditioning/accuracy.

Base Case Injection Average Steps

Number of time steps at the end to average over to calculate base case injection rates for prediction.

Base Case Time Steps

Number of default time steps in predict input file, used to run what-if scenarios.

3.1.11.5 Well Multiplier

Apply Multipliers

Multipliers can be used to refine fit for individual wells. The multipliers increase or decrease proportionally the productivity or injectivity of a well. The default value is 1 when not specified.

3.1.11.6 Polymer

This section displays parameters applicable only to the Polymer Injection model type.

Reservoir Initial Polymer Conc.

Initial Polymer Concentration [ppm]

Skin Per Log Cum Polymer wt

Skin Factor log growth rate per cumulative production.

Maximum Skin Factor

Maximum Skin Factor.

Salt Density in wt/(bbl*conc)

Salt density per solution concentration wt/(conc*bbl).

Polymer Density in wt/(bbl*conc)

Polymer density per solution concentration wt/(conc*bbl).

Polymer Adsorption K

Polymer Langmuir Adsorption Equilibrium Constant

Polymer Adsorption Max

Polymer Langmuir Maximum Adsorption.

Polymer Viscosity Coeff. (A)

Polymer Coefficient for Water Viscosity (A).

Polymer Viscosity Coeff. (B)

Polymer Coefficient for Water Viscosity (B).

Polymer Viscosity Coeff. (C)

Polymer Coefficient for Water Viscosity (C).

Polymer Viscosity Coeff. (N)

Polymer Coefficient for Water Viscosity (N).

3.1.11.7 Uncertainties

Water Injector Well Index Multiplier

This multiplier for water injection is required so the same reservoir responsivity map can be used for both injectors and producers.

Gas Injector Well Index Multiplier

This multiplier for gas injection is required so the same reservoir responsivity map can be used for both injectors and producers.

Additional Pressure Differential

Pressure head at the producer due to fluid column and/or artificial lift. This parameter is actually additional energy inside the reservoir that may not be captured by “Pr-BHP”. For steam-flooded reservoirs, this parameter is needed. For waterflooded / WAG / Primary reservoirs, this term is very small (a few psi) and is not significant.

Pressure Differential Decline Rate

Rate of reduction of dpLift, larger this is faster the gross and oil rate decline.

Oil Viscosity Rate of Change

Defines the rate of change of oil viscosity with pressure, the larger this number, the faster the change. Data Physics models solve for a viscosity-pressure relationship as viscosity is a function of pressure, and as pressure is changing in reservoirs. This parameter defines the rate of change of the oil viscosity due to pressure. A higher value results in viscosity changing faster as a function of pressure [cP/psi].

Gas Viscosity Rate of Change

Defines the rate of change of gas viscosity with pressure, the larger this number, the faster the change. Data Physics models solve for a viscosity-pressure relationship as viscosity is a function of pressure, and as pressure is changing in reservoirs. This parameter defines the rate of change of the oil viscosity due to pressure. A higher value results in viscosity changing faster as a function of pressure [cP/psi].

3.1.11.7 Constants

Water Viscosity Rate of Change

Defines the rate of change of water viscosity with pressure, the larger this number, the faster the change. [cP/psi].

Molar Weight of Gas

Gas Density [lb/ft^3].

Reference Non Ideal Gas Factor

Z factor of free gas at reference pressure.

Non Ideal Gas Factor

Z factor of free gas at reservoir condition.

Pressure of Non Ideal Gas Factor

Pressure at which Z factor of free gas is given, normally at reservoir condition. Default 0 assumes it applies constant value of ZGas regardless of pressure.

3.1.11.8 Fault Parameters

3.2 MODEL RESULTS

Once a model (either Back Test or Full Fit) has successfully run, results become accessible by clicking the case name within the Models section list.

Figure 28. Model Results Display.

Top bar buttons:

Model Configuration

Displays all configured model parameters, including overview, general setup, well group, fit setup, solver controls, fit bounds, parameter ranges, and constants.

Run

Re-executes the model based on existing or modified configurations. Available options:

• Run. Rerun the model, overriding the current version.
• Run as New Model. Execute the model using a new name and description.
• Run as New/Updated Model. Reevaluate and update model configuration step by step.

Dashboard

Leverage external dashboards to compare different scenarios and conduct a more thorough analysis of the results. These dashboards enable the review and plotting of wells with highly configurable options, facilitating the swift identification of wells/layers with the most significant changes.

Scenarios

Enables a window displaying all scenarios created for the selected model and provides the option to run a new scenario.

Bottom tabs:

3.2.1 Model Diagnostics

Figure 29. Model Diagnostics.

Wells evaluated

The first dropdown list enables to review the results sorted by All Wells, New Wells (not included in training) and Old Wells (included in training).

Phase

Results are allowed to be displayed for all producing and injecting phases: oil, gross, water, gas, water, gas or polymer injection and produced polymer concentration.

Coefficients Chart

A table is displayed in the center of the window with the results of Spearman and Pearson correlation coefficient factors calculated for the selected phase. Both All Wells, Old Wells, and New Wells coefficients are evaluated. Spearman and Pearson coefficients are calculated for the real vs. predicted cumulative productions plotted on the crossplot chart and are used to validate predictive capacity at the well or well-layer level. It is expected to obtain higher coefficients for the old wells than for the new wells since the latter has not been incorporated into the model for training.

Training Graph (Field Level)

Visually represent the ensemble’s training per time step throughout the time-lapse designates and the validation period, where the model demonstrated how well the training went. The graphs are constituted by:

• The red area signifies the learning period, during which the algorithm assimilates information from the production history, culminating in the determination of global parameters. Each data point symbolizes an ensemble assessing the initially defined range of properties for the model. Over time, the convergence of these properties is refined through the application of ensemble Kalman filters.
• The interception denotes the point at which the algorithm ceases its learning process.
• The blue area represents the subsequent prediction period, founded on the knowledge acquired during the learning phase. Each line corresponds to a distinct percentile, with the central thick line depicting the mean field behavior.

Actual vs Calculated Crossplot (Well Level)

The display on the right shows a comparison between calculated cumulative production vs actual cumulative production at individual wells, for the prediction period. It is desired to obtain coefficients above 0.6 for both Spearman and Pearson correlations and to have a good trend between the actual data measured in the field and the mean of the ensembles. Both Spearman and Pearson correlation coefficients measure the linear relationship between predicted cumulative production on the Y-axis and measured cumulative production on the X-axis.

3.2.2 Parameters Sensitivity

Figure 30. Parameters Sensitivity.

A Tornado chart compares the relative impact of the main parameters for each fitting phase. This chart allows the user to adjust parameters during the fitting phase, where the parameters are adjusted and iterated to create a good model. If each variable is selected below (Injection, Gas, Liquid, Oil), only the corresponding bars will be shown in the Tornado chart.

3.2.3 Responsivity

Figure 31.Responsivity.

The plot shows the mean responsivity at both the field and well levels, evaluated over the time series. The responsivity changes with each time step during the training period but remains constant during the testing or validation period.

Filters (only displayed in the Responsivity tab)

When this option is enabled, editable options for prediction scenarios are displayed to configure charts:

• Options
  • Measurements. Allows graph visualization for Liquid Responsivity, Oil Responsivity, Gas Responsivity, Historical Pressure, Water Saturation and Gas Saturation.
  • Scale. Linear or Logarithmic scale to plot the information.
• Filter Wells.
  • Select Wells. Permit select or unselect a group of wells.

Figure 32. Wells Filter.

• Search bar. The search bar allows you to quickly find the relevant wells within the list.
• Select All Wells. The Wells table enables the selection of all the wells at the same time by using the checkbox next to the wells name.
• Filters button. Wells associated with a certain reservoir (in the case of the reservoir model), Fault Block, or Compartment are selected automatically through this option.
• Wells List Chart. Selection of all or a group of wells (injectors or injectors and Producers) desire to be targeted for results visualization.
• Your Selection Chart. Displays the chosen well selection that will be taken into account for plotting the charts.

3.2.4 Histograms

Figure 33. Histograms.

The histograms shown in this section corresponding to the physical parameters that yield forecasts that match the observed behaviors while honoring physical laws. These histograms help to understand that the solution not only reproduces the historical production trend but also represents the final ranges for the tuning parameters within the initially given ranges.

As part of the validation process, histograms for the various reservoir parameters need to be validated to confirm they are within reasonable ranges. Data Physics models search between user-specified bounds for combinations of physical parameters that yield forecasts that match with observed behavior and honor physical laws. The resulting parameter derivations are non-unique, and both the underlying system-state maps and histograms will not exactly match static models or other interpretations that E&P companies typically perform. They are displayed above as a validation to reveal whether a particular fit deviates significantly from conventional understanding. Properties measured at the core and log scale are not necessarily the most appropriate to use when considering larger scales in which the models are built.

Histograms expose in the Y axis the number of ensembles using a specific property value and on the X axis is the user-defined property range. The blue color represents the distributions at the beginning of the simulation (randomly initialized) and the green color refers to the final distribution obtained at the end of the learning period and used in all future forecasts.

3.2.5 Responsivity Maps

Figure 34. Responsivity Map.

This section denotes the different underlying maps that are fitted and used in creating the individual well’s production forecasts. Pressure, Water Saturation, Gas Saturation, and Oil Saturation maps are estimated, as well as the Responsivity maps for Liquid and Oil, which represent the quality of the reservoir. These Responsivity maps combine K and Well Index, where Well Index (WI) is essentially the geometric well factor per unit completion length used in the well model in reservoir simulators. Responsivity maps can also be described as the productivity well index multiplier applied among wells and characterized continuously over space.

Within the Responsivity maps window, all the different properties maps can be selected, as well as the different domains (reservoirs, fault blocks). All of the above-mentioned properties will correspond to the last timestep used in the training. For each property selected, the information icon will show a description of it. The two drop-down lists available within the window allows to select the map (property) and unit (layer/fault block or compartment).

3.2.6 Files

Figure 35. Files Segment.

Generated files in the fit are exposed in this tab; three different file sections are shown: input, output, and output-csv. The files included in this section are utilized by the software to plot results and create tables.