ComboCurve: Generating Automatic Deterministic and Probabilistic Forecasts Updated

ComboCurve: Generating Automatic Deterministic and Probabilistic Forecasts Updated

Overview

The automated forecasting engine conducts a regression analysis on the production data of a specified well, generating a fit based on the configurations provided in the auto-forecast settings. Access to this feature is available by initiating a new forecast and selecting the "Run Forecast" button, prominently displayed at the top middle of the interface.

Forecast Settings

Parameter Set

The configurations within the Forecast Settings are critical in determining the outcome of the analysis. A Parameter Set encapsulates a series of configurations applicable to one or several phases of well operations. For uniform application across all phases, a single parameter set suffices. Conversely, to apply distinct configurations to different phases, multiple parameter sets are required, as illustrated below:
  1. Option 1: Uniform Settings Across All Phases
    1. By default, the same forecasting configurations are applied across all phases. To omit a specific phase from the forecast, one may deselect it by clicking the adjacent 'x'. For instance, to forecast only oil and gas while excluding water, simply deselect the water phase. This principle is applicable universally across all phases.

  1. Option 2: Unique Settings for Individual Phases
    1. To assign unique settings to each phase, the number of parameter sets should equal the number of distinct configurations required. For example, to forecast oil, gas, and water under different settings, three separate parameter sets should be created, each assigned to its respective phase as depicted.

Note: Additionally, Custom Streams facilitate the forecasting of phases beyond Oil, Gas, and Water. By navigating to "Forecast Options" and selecting "Stream Management", users can augment their forecast with additional streams.

Forecast Resolution

This segment defines the granularity of production data utilized in the forecast, with options including:
  1. Monthly Only
  2. Daily Only
  3. Monthly Preference: defaulting to daily if unavailable
  4. Daily Preference: defaulting to monthly if unavailable

Model Selection and Model Field

  1. Model Selection outlines the forecast model applied to the analysis.
    1. Rate: Forecasts each phase independently.
    2. Ratio: A phase's forecast depends on a base phase, necessitating at least one phase to be forecasted as a rate to model others as a ratio.
  1. Model Name offers a selection of Decline Curve Analysis (DCA) models.
  1. Model Fields include the necessary inputs depending upon your choice of forecasting model.  For example, if choosing Modified Arps as the forecast model, the necessary fields will be as follows:
    1. b factor: b factor range for the Arps decline
    2. Di-Eff-Sec (%): initial secant effective decline (%/year)
    3. D Sw-Eff-Sec (%): effective terminal decline at which the Arps segment switches to exponential
    4. Enforce Switch: enforces switch to exponential decline even if that occurs before the end of production data
    5. Well Life: number of years from selected date
    6. q final: terminal rate
    7. Peak Preference: The point at which the curve hangs from, this can be:
      1. Beginning Point
      2. Last Peak
      3. Max
      4. End Flat Point
      5. Automatic
      6. Note: you can hover over the tooltip icon on each option to see a brief explanation of each one
    8. Zero Forecast (Days from Today): If there is no production data from that number of days in the past to today, the forecast will be shut in
      1. Example: Today's date is 4/9/24 and the value is 365, if there is no production data from 4/9/23, the forecast segment will be a shut in
Note: Not all models will require the same input fields. For example, if choosing exponential decline, there won't be a requirement range for b factor since it's already set to 0.

Filter Options

These settings allow for the exclusion of outliers and customization of the data set for regression analysis.
  1. Data Density: filter to remove outliers, the higher the level, the more outliers will be excluded
  2. Date: this represents the portion of production data you want the forecast to be based on. You can leave this to "All' to include all production data or you can select a specific time period such as "Last" 1-x months
  3. Value: this excludes all production rates that go above or below this interval (measured in bbl/day for oil/water and mcf/day for gas)
  4. Percentile: this excludes all production rates that go above or below this percentile
  5. Moving Average: production is averaged for a selected number of days to generate a smoother fit

Weighting Data

This functionality enables the adjustment of weighting applied to specific data segments, influencing the forecast's emphasis.
  1. A weight value of 1 represents no difference
  2. A weight value of 10 puts more weight on the selected range compared to the remaining data
  3. A weight value of 0.1 puts more weight on the remaining data compared to the selected range

EUR Restrictions 

These options align the forecast with specific Expected Ultimate Recovery (EUR) values:
  1. Match Forecast: matches the forecast EUR to the same well on a different forecast within the same project
  2. Fixed Value: matches the forecast EUR to a specific value

Low Data Forecast Settings

  1. Require Minimum Amount of Data to Forecast: Sets a threshold for the volume of data necessary to initiate the forecast.
  2. Enable Historically Informed Forecast: Utilizes historical data trends when insufficient recent data is available.

Overwrite Manual

Toggles the ability to override manually edited well forecasts upon executing an automated forecast. If the setting is left uncheck, running the auto forecast will not override the manual forecast. On the opposite, if left checked, the run will override every well in the forecast. 

Troubleshooting Possible Errors

In instances where wells do not meet the forecasting algorithm's prerequisites, identifying and addressing warnings post-forecast is advisable. To filter down to wells with warnings, you can click on Forecast Filter >> Show Wells with Warning. 

 After filtering, you can see all of the wells that did not run. You can click on the yellow triangle next to the forecast chart of each individual well to get a brief description of the warning to better understand the issue. 


In this example, for the oil forecast, we selected a Peak Preference of "Last Peak". For this specific production profile, the last oil peak is located a the last production datapoint. This would cause the forecast to not run because there is not enough data after the peak to generate a fit as the error message describes. To fix this issue, you can re-run the forecast for just this well with slightly different settings. To do so, you can click on the 3 dots right above the yellow triangle and select Single Well Forecast. This configuration will only apply to the specific well and will not alter any of the other wells' forecasts. For this example, we can change the peak rate from Last Peak to Max to provide enough data to generate a fit. 


Alternatively, you can check the box to the left of the well chart to add the well to the editing bucket. Once in the editing bucket, you can click on the Editing tab at the top of the screen to manually forecast this well.

    • Related Articles

    • ComboCurve: Custom Streams Workflow

      Overview This article presents a comprehensive workflow that encompasses the creation and utilization of custom streams. Specifically, it outlines the process of importing production data, running forecasts, and conducting economic analyses using ...

    • ComboCurve: Type Curve Methodology Documentation

      TC Settings Associated Forecast TC needs an associated forecast to generate automatic type curve fits and automatic normalization fits All the wells in the TC must be in the associated forecast as well, TC wells is a subset of forecast wells TC can ...

    • ComboCurve: Auto Proximity

      What is Auto Proximity? Auto Proximity allows you to run proximity forecasting for multiple wells at once. As a side note, proximity forecasting consists of generating a localized type curve in which the analog wells are selected based on location ...

    • ComboCurve: Company Custom Headers and Company Custom Streams

      Company Custom Headers Company Custom headers can be created at the company level under the custom columns tab. Inside the custom columns tab you will see that two options, well headers and production steams. Clicking on well headers allows the user ...

    • ComboCurve: Review Forecasts in Scenario Page

      ComboCurve: Review Forecasts in Scenario Page In this article we will discuss how to select your forecasts for wells, along with being able to edit or view them one last time. Choose Forecast Once a forecast has been created for all of the wells in ...