Oil Sands Well Interaction Detection
Industry
Use Case Business Improvement

Challenge

  • In oil sands, on adjacent wells, there is potential for materials to move between wells (“well interaction”)
  • Goal is to detect and take action on well communication when it is:
    • Highly desirable – CO2 Enhanced Oil Recovery (EOR), Steam-Assisted, Gravity Driven (SAGD) in oil sands
    • Or undesirable – introducing frack material or stimulant into production wells, or water from poor performing wells to others
  • Needed on hundreds of pads, and 1000s of wells

 

Solution

  • Detect: sharp changes in pressure, temperature, conductivity
  • Seeq calculation engine used for regression, and signal time offsets
  • Using Seeq’s Matlab SDK, systematically search across hundreds of well pairs to find correlations
  • Asset structure includes assignment of “neighbor wells”

 

Benefits

  • Identify “offender” wells with undesirable well communication in a SAGD field
  • Product quality improvement and reduced processing costs
  • Reduce damage SAGD well liners due to steam infiltration
  • Energy and material savings (Steam, CO2, chemicals)
  • Offset costly consulting services

 

Data Sources

  • Process Data Historian: OSIsoft PI
  • Asset Structure: OSIsoft Asset Framework

An AF structure linking “neighbor” wells is developed for systematic detection of relationships between process conditions in wells and their neighbors.

 

Data Cleansing

  • Signals are shifted in time to account for delay and residence time of material migration from “perpetrator” to “victim” wells

 

Calculations and Capsules

  • Relationships between neighbor wells are identified with OLS1 regression analysis of steam injection in “perpetrator” wells and temperature in “victim” wells
  • To account for migration delays, signals are shifted in time to identify the true relationship of migration
  • Leveraging the Matlab SDK, temperature signals in “victim” wells are systematically shifted 4 hours to 4 days and correlated with steam injection rate of the “perpetrator” wells (Correlation coefficients indicate relevance and rank of importance for review prioritization)

 

Summarizing Results

  • Migration time between “perpetrator” and “victim” wells identified as typically averaging 8 hours, where present
  • Well interaction consistently identified when R2 of steam injection/victim well temp (shifted 8 hours) was 0.2 or higher
  • Identification of EOR2 material migration can be systematically affiliated with other process occurrences, such as liner degradation or well productivity anomaly data that is overlaid