Intelligent digital platform for Oil Reservoir



Challenge Summary

An intelligent platform for oil reservoirs to implement new modelling approach, which combines data, physics, chemistry and learns from field observations, plan, prioritise, assess potential well operational activities in order to achieve higher operational efficiency, production rates from the reservoir, incorporation of testing mechanism for hypothesis and find candidates to meet production goals.

Challenge Scenario

The reservoir management operates at the confluence of technical, operating, and managerial skillset. It is a continual process that recognizes the limitation and uncertainties in the modelling of complex reservoir operations. The traditional approach of reservoir modelling is based on first principle physics with partial differential equation and represented by a geo-cellular (static) model, developed by geoscientists and is dominated by interpretations and uncertain values. The inability of E&P companies to reduce the uncertainties in reservoir management, pose a significant risk to the identification of the right candidate for advanced oil recovery techniques such as EOR/IOR, optimal flooding pattern and RoI analysis prior to full-scale implementation, resulting in loss of hydrocarbon recovery potential. A system for oil reservoirs to implement new modelling approach, which combines data, physics, chemistry and learns from field observations, plan, prioritise, assess potential well operational activities is required.

Profile of the End-User

The potential users of the proposed system would be:

1. Reservoir Engineers
2. Production Technologists
3. Petrophysicists
4. Production chemists
5. Well Services Specialist

Any specialist who would like to assess the effectiveness of chemical treatment in the well bore, evaluate different hypotheses on small scale laboratory experiment simulations and upscale to the field scale (e.g. formation damage mechanism on lab.scale, near well bore zone and deep in the reservoir)Existing approaches are narrowly discipline focused, do not allow to solve this type of the problems in combination with reservoir model.

There is a lot of subsurface and surface data, but the data is not consistent enough to build high quality robust reservoir models in tools like CMG and Eclipse in short time. The existing tools are difficult to use for daily planning of activities and at the same time include all the recent available data. Despite the presence of the calculation results in existing 3D models, the functionality of the post processing of simulation data is not sufficient to provide recommendations what to do on the daily basis with the well and reservoir. A new modeling approach, which combines data, physics, chemistry is required to learn from field observation and build high speed reservoir predictive models. The approach must reflect existing field complications: polymers, deposition and dissolution processes and be robust enough to model probabilistic scenarios for field areas up to 100 wells.

Functional Requirements of the End-User

  • Upload existing simulation model from CMG, Eclipse
  • Upload large volume of historical data
  • Maximise the Value of Information VOI
  • Implement the new physics/chemistry mechanism to be simulated
  • Quick and automatic HM 1-2 hr.
  • Check of quality of HM
  • Estimate and Export from simulation model the data useful for well operations such as ranked list of the wells and interval candidates recommended for specific treatment with all necessary statistics for the well, interval, flow channel.

Functional & Operational Capabilities

  • High simulation speed, 3 years field forecast run under 15 min
  • Reflect polymers, waxes, asphaltenes, plugging processes, water-oil emulsion design, tracers, temperature simulation
  • Easy-to-use front end
  • Easy to operate using scripting languages (python and so on)
  • Local or cloud computations
  • Extraction of statistics based on well, flow-channel, intervals
  • Estimation the displacement efficiency of well/ flow-channel
  • Ranking the wells/flow-channels for different types of well operations
  • Forecasting with "simulator best recommended operations" and manually selected operations

Operational constraints

  • Easy to deploy in any public or private cloud environment
  • Availability of data in a structured manner
  • Data security
  • Migration of data from existing platforms
  • Existing infrastructure to gather various data of well parameters.

Expected Tangible Benefits and Measurable Gains

  • Mitigate production decline 3% less
  • Increasing the Value of Existing Information VOI by incorporating significant more data amount for daily operations, comparing to the situation now
  • Selection of best/fit for purpose candidates to achieve production target
  • Increasing the reserves and sweep efficiency
  • Increasing VOI and assurance of future reservoir performance by incorporation and testing mechanism of different hypotheses

Performance Metrics or Outcomes

  • Quality of automated history matching, 12 years history
  • Comparison actual well, reservoir performance vs simulation results: 1 - 36 months forecast
  • Matching with laboratory experiments

​​​