Production Forecast (DCA) Package

Perform automated multi-pass decline curve analysis (DCA) with outlier identification and forecasting. List of Fit Codes.

Usage

Run the workflows from this package to perform the calculations. The schedules may be adapted to use different contexts.

Overview

The Production Forecast (DCA) package performs automated Arps decline curve analysis (DCA) using a multi-pass approach to fit and forecast production. Forecasts are saved to the scenario Production Forecasts (DCA).

The multi-pass approach is used to identify influential outliers and refine the final fit of the production data. Once the final fit is created, the package will forecast all producing phases (oil, gas, condensate and water) for a configurable time span starting at the end of an entities historical production.

Fits created by the package will be named AutoDCA xxx Fit​, where xxx is one of Oil, Gas, Condensate or Water.

Details

Workflows

The package uses the following workflows, in sequence:

Production Forecast (DCA): Reset​ - deletes fits and clears related signals

Production Forecast (DCA): Assign Fit Codes - determines the fit code for each fitted phase. Note that only fit codes > 0 will be considered for fitting and/or forecasting. See the section Fit Codes below for an explanation of the valid fit codes

Production Forecast (DCA): 3-Parameter Fit (First Pass) - performs a first pass three-parameter fit (Qi, Di, bexp) using all available data of the specified fitting phases (see workspace value AutoDCAPhasesToFit) that have an assigned three-parameter fit code (fit code = 3)

Production Forecast (DCA): 3-Parameter Fit (Errors And Outliers) - identifies outlier points based on the first pass fit

Production Forecast (DCA): 3-Parameter Fit (Second Pass) - performs a second three-parameter fit excluding outlier points from the analysis

Production Forecast (DCA): Re-Fit Evaluation - fit codes are reassigned as necessary based on rules using the fitting phase IP90 KPI, EUR and cumulative production

Production Forecast (DCA): 2-Parameter Fit (First Pass) - performs a first pass two-parameter fit (Qi, Di) using all available data of the specified fitting phase (see workspace value AutoDCAPhasesToFit) that have an assigned two-parameter fit code (fit code = 2 or 5)

Production Forecast (DCA): 2-Parameter Fit (Errors And Outliers) - identifies outlier points based on the first pass fit

Production Forecast (DCA): 2-Parameter Fit (Second Pass) - performs a second two-parameter fit excluding outlier points from the analysis

Production Forecast (DCA): Post-Processing - calculates fit and forecast rates for phases with a three- or two-parameter fit and default forecast rates for phases that have an assigned default fit code (fit code = 1, 6, 7 or 8). Also calculates secondary forecast rates

Production Forecast: Forecast Volumes - calculates forecast volumes from the forecasted rates

Workspace Values

AutoDCAForecastMonths​ - defines the number of months to forecast starting at the end of an entities historic production data

AutoDCAOfflineTimeSteps​ - sets the number of trailing time steps from the end of an entities historic production that defines an entity as being offline. Offline entities are excluded from the analysis

AutoDCAOutlierMethod​​ - defines the outlier detection method used to identify influential outliers in the fitted production data. All methods use fit error deviation as the basis for detecting outliers. There are three options (string values; not case sensitive) for this workspace value:

• Option 1: linear - uses a simple difference between an observed data point and the fitted value for that data point when calculating the error at that point. That error is then compared to the RMSE of the fit
• Option 2: log - uses a difference of the natural logarithms of an observed data point and the fitted value for that data point to define the error at that point. That error is then compared to the RMSLE of the fit
• Option 3: both - calculates the error using both linear and natural logarithm differences for each data point and compares those errors to the RMSE and RMSLE, respectively, of the fit. To be an outlier, both the linear and the natural logarithm error/RMSE ratios must exceed AutoDCAOutlierThreshold

​AutoDCAOutlierThreshold​ - sets the magnitude of the ratio (absolute error)/(Root Mean Square Error (RMSE)) that defines an outlier data point. Ratios of the errors calculated by AutoDCAOutlierMethod to the RMSE that exceed AutoDCAOutlierThreshold are considered outliers. The default value is 1.5

AutoDCAPhasesToFit​ - defines the phases that will be fit using the workflow described above. There are two options (string value; not case sensitive) for this workspace value:

• Option 1: primary - with this option, only the primary phase is fit with AutoDCA. The primary phase is set using the Assign Primary Phase P# script (package Basic). Use AutoDCAPhasesToFit = primary if you only want the primary phase to be fitted in your project. Phases that are not the primary phase will be forecast using trailing 6-months average fluid ratios relative to the primary phase of the entity. See also workspace value PhasesInWorkspace
• Option 2: any combination of oil, gas, condensate, and water - with this option, each phase in the list is independently fit with AutoDCA. Other phases that are not in the list are forecast using trailling 6-months average fluid ratios relative to the primary phase of the entity. See also workspace value PhasesInWorkspace

AutoDCARegressionWindow​ - sets the size of the regression window used to define what data should be evaluated for fitting from the end of an entities production rate signal for a phase.

MinCondensateProductionRate​ - sets the minimum flowrate for condensate that would be considered in an analysis. Rates less than or equal to the set value are excluded from the analysis. The default is 0.0

MinGasProductionRate​ - sets the minimum flowrate for gas that would be considered in an analysis. Rates less than or equal to the set value are excluded from the analysis. The default is 0.0

MinOilProductionRate​ - sets the minimum flowrate for oil that would be considered in an analysis. Rates less than or equal to the set value are excluded from the analysis. The default is 0.0

MinWaterProductionRate​ - sets the minimum flowrate for water that would be considered in an analysis. Rates less than or equal to the set value are excluded from the analysis. The default is 0.0

ProducedPhasesInWorkspace​ - defines a list of all the phases that exist in the database. ProducedPhasesInWorkspace can be any combination of oil, gas, condensate, and water. In the workflow, some scripts use this list to enumerate through each phase to perform the calculations

AutoDCATimeIncrement - sets the time increment for retrieving production data for the DCA analysis. It is also used to define the step used as regression window (see AutoDCARegressionWindow). E.g. AutoDCATimeIncrement = Monthly and AutoDCARegressionWindow = 12 -> the regression window will consist of the last 12 months before the end of an entities production data

AutoDCAMinRateOil - sets the minimum flow-rate for the entity's oil forecast.

AutoDCAMinRateGas - sets the minimum flow-rate for the entity's gas forecast.

AutoDCAMinRateWater - sets the minimum flow-rate for the entity's water forecast.

AutoDCAMinRateCondensate - sets the minimum flow-rate for the entity's condensate forecast.

AutoDCAMaximumConstant - sets the maximum value for the Arps hyperbolic decline constant (bexp) for a hyperbolic fit.

AutoDCAMminimumConstant - sets the minimum value for the Arps hyperbolic decline constant (bexp) for a hyperbolic fit.

AutoDCAMaximumSlope - sets the maximum value for an initial Arps decline rate (Di) for a fit. The entered value should be an annual rate in #/year. For example, a 6% annual decline rate should be entered as 0.06.

AutoDCAMinimumSlope - sets the maximum value for an initial Arps decline rate (Di) for a fit. The entered value should be an annual rate in #/year. For example, a 6% annual decline rate should be entered as 0.06.

AutoDCATerminalDecline - sets the terminal decline rate at which a hyperbolic or harmonic curve should transition to an exponential decline. The entered value should be an annual rate in #/year. For example, a 6% annual decline rate should be entered as 0.06.

AutoDCAReviewThreshold - sets the R-squared value below which a fit will be flagged for manual review.

AutoDCAMinimumGapSteps - sets the minimum allowable number of time steps with no production data before truncating the regressing window. If a gap larger than this minimum is detected inside the regression window, the regression will be performed from the end of this gap forward.

Placeholders

​Please specify a phase...​ - this defines the phase a script is executed for. Must be one of Oil, Gas, Condensate, Water.

​Max bexp​ - maximum hyperbolic constant; used during re-fit evaluation to check for excessive EURs

​Default bexp​ - default value for the hyperbolic constant used for constructing default forecasts (fit codes 1, 6, 7, and 8)

Default Dmin​ - default value for the minimum decline rate used for constructing default forecasts (fit codes 1, 6, 7, and 8)

Default Di​ - default value for the initial decline rate used for constructing default forecasts (fit codes 1, 6, 7, and 8)

Please specify a mode... - the mode for calculating volumes from rates. Depending on the selection either the average uptime of the last six months, a fixed (user defined) value, or the number of days at the time step is used

Fit Codes

The following fit codes are used in this package. Note that only fit codes > 0 will be considered for regression and forecasting.

• -7: the rates inside the regression window are less than the workspace value threshold MinXXXProductionRate (XXX = Oil, Gas, Condensate or Water)
• -5: all the regression data became outliers during the calculations
• -3: entity has been offline too long; has multiple criteria
• -2: primary phase is NO PRODUCTION and there is no production in the database
• -1: primary phase is UNKNOWN (meaning, it could not be identified as Oil, Gas, Condensate or Water)
• 0: there was an error during the analysis
• 1: entity only has one production data point; triggers a default forecast construction using the available data point as initial rate, and default values for the initial and minimum decline rate and hyperbolic constant
• 2: less than six and more than two production data points; triggers a 2-parameter fit using a default hyperbolic constant
• 3: sufficient data for a full 3-parameter fit
• 4: the forecast type is something other than automated regressed Arps (like manual or user defined)
• 5: EUR/IP90 violation, indicating possible excessive EUR relative to the best three-month production rate (IP90); triggers a 2-parameter re-fit. Conditions for a fit code of 5 are:
  • cumulative production ≤ 0.2 * EUR
  • EUR/IP90 ≥ 2.0E03
  • hyperbolic constant ≈ maximum hyperbolic constant
• 6: curtailed production (relatively flat or low-decline production profile); triggers a 1-parameter default production forecast curve construction using the last flow rate as the starting point for the forecast. Conditions for a fit code of 6 are:
  • cumulative production ≤ 0.2 * EUR
  • EUR/IP90 ≥ 4.0E03
  • Di/Df ≤ 3.0 (relatively flat or low decline rate)
• 7: the maximum flow rate is the last production data point; triggers a 1-parameter default production forecast using the last data point as initial rate and default values for the initial and minimum decline rate and hyperbolic constant
• 8: only two data points to fit; triggers a default production forecast construction using the two available points to define a slope, the first data point for the initial rate and a default hyperbolic constant and minimum decline rate

Data Requirements

The following data is required:

production rates (oil, gas, condensate, water)
production volumes (oil, gas, condensate, water)
tag entries for the tags Active, Oil, Gas, Condensate, Water

Further Comments

Manually created/edited DCA models will not be deleted or updated by the workflows. However, forecasts will be calculated for all available DCA models. To recalculate the forecasts of a manual DCA model after saving it, run the workflow Production Forecast (DCA): Refresh Manual Fits. If a manual fit shall be overridden by the workflows, delete it before starting with ​Production Forecast (DCA): Reset.