noun_Email_707352 noun_917542_cc noun_Globe_1168332 Map point Play Untitled Retweet

Join us at SSV - energy:connected conference 2019

Take the opportunity to hear us speak about how oil well performance can be improved by help of predictive maintenance, supported by Artificial Intelligence.

10 - 11.4.2019, Norway

Tieto presents experiences from a pilot project addressing challenges of improving oil well performance by help of predictive analytics of Electrical Submersible Pumps at this year’s energy:connected conference @ Fornebu 10th of April.

Join us! Our session starts at 11 am at the center pitch stage - read more here.

Scope of this project was to address the challenges of improving Electrical Submersible Pumps (ESP) operations uptime and performance through Artificial Intelligence-powered predictive and prescriptive maintenance.

More specifically, three main ESP failure risks (Pump wear riskBroken shaft risk and Resonance vibration risk) were calculated based on pressure and vibration sensor data.

Value unlocked through the project:

  • It was proved that sufficient data was present to create predictive models for the pumps uptime and also performance.
  • It was clarified that predictive performance with risk weighting was a viable methodology for production planning.
  • The interactive dashboard was proven to be useful for all users, i.e. data scientists, operators, production planners and the internal maintenance teams.
  • Predictive modelling was also proven via the advanced scenario tool.
  • The Analytical models were made tunable by the User thru the User Interface (self-service predictive modelling).

 

ArtificiaI Intelligence integration scenarios:

  • Real-time AI integration for predictions (e.g. every 100ms to 1m sampling interval)
  • Batch processing of data for model training (e.g. deep learning, re-enforcement learning, model training)
  • AI model health checks (continuous validation of AI predictions

System architecture consists of:

  • Tieto Service Connect IoT accelerator
  • Microsoft Azure Cloud Platform
  • PowerBI Visualization
  • Python Scikit-Learn
  • Python FLASK

Contact us

Helene Helvik

Sales Director

Dato og sted

10 - 11.4.2019
Del på Facebook Tweet Del på LinkedIn