Energy & Utilities

To offset the fluctuations in the cost of oil, the energy industry needs improved efficiencies in all parts of its production and delivery chain. Predictive analytics leverages the large volumes and varieties of data to discover critical patterns capable of improving performance, and reducing losses and downtime.

With the rise of data capture and digital transformation in the energy industry, there is increased opportunities to make data-driven decisions along the production and delivery chain. Elder Research is a leading designer and implementer of cutting-edge advanced analytics solutions in the energy sector.

Our energy analytics solutions include:

  • Improved Demand Forecasts
  • Optimized Energy Trade and Spend
  • Enhanced Predictive and Preventative Maintenance

Request a Consultation

Contact Us

Case Studies

We help you understand and use your data assets to reveal advanced insights leading to improved decision-making, optimal asset utilization, and reduced costs. Examples of our energy solutions include:


Predictive Maintenance of Hydroelectric Generators and Turbines

We helped Sira-Kvina kraftselskap, one of the largest hydro-power producers in Norway, build a predictive maintenance framework for finding component failures in generator and turbine systems. The team used a finite state machine (Ceptor) to extract insights from over 1400 sensors across 16 facilities, then matched those against several years of maintenance records and inspection reports to build survival (Cox), deep learning (LSTM), Bayesian, and classifier (GBM) models of system performance.  Results were fed into a visualization tool (RADR) to highlight risks for the engineering team.

Results:  The resulting framework was able to identify the cause of a particularly elusive generator failure that had stumped the maintenance team.  Finding the source of this fault was enormously important, because this generator produces tens of millions of euros per month in electricity, making any down-time particularly costly.  Read the case study


Using Sensor Analytics to Predict Natural Gas Well Failure

We helped an international oil and gas exploration firm by harnessing 20 years of detailed (but very noisy) sensor readings from hundreds of wells to characterize transient well states. The client needed to predict gas well shut-ins (blockages preventing production) 4 to 6 months in advance to effectively mitigate risks.

Results: Our data science model was far more accurate than the existing baseline models. Our client now had predictions of clusters of well pads that should be prioritized for treatment or preventative maintenance. Payback for this engagement was one year. Read the Case Study