Shifting from Reactive to Predictive in Energy: Driving Valuable Outcomes in a New Digital Landscape Part 1

Technology and Digital


May 16, 2018

For the Oil and Gas industry to succeed in making the digital shift from reactive to predictive maintenance, their processes and technology applications need to evolve. In this three-part blog series, we will discuss the following:

  • Incorporating Data Analysis into Operations Processes, Part 1
  • Human and IT impacts, Part 2
  • Managing Digitalization Programs, Part 3

Most digital projects today rely heavily on disparate data generated from censored, physical objects or “things” – as used in the common term, Industrial Internet of Things (IIoT). To put some numbers on this, Gartner, Inc. forecasted that 8.4 billion connected things would be in use, worldwide by 2017, up 31 percent from 2016, and will reach 20.4 billion by 2020. The number of individual sensors could top one trillion as of 2020 according to estimates from several companies and industry groups. As the number of installed, object-level sensors increase, vast quantities of new data are sent, stored and must be analyzed to build predictive models and prevent break down on equipment; this is accomplished by developing and training predictive algorithms in the form of neural networks and implementing those algorithms within the appropriate technology-based solution frameworks.

Effective data management requires that the implementation of digital technology considers requirements for cyber security, infrastructure and operational information capture, with minimal disruption to normal operations. From this point, the development of data analysis routines and models will help to build more predictive algorithms. For example, data scientists can identify patterns that enable operators to be notified that a piece of equipment requires service when time series performance data indicates a given operational process parameter such as pressure, temperature, vibration, etc., is falling outside of normal operating limits; providing an early warning signal that a given equipment object may be experiencing health issues or even approaching failure.

Another common industry example focuses on the Owner Operator’s goal to keep operations up-and-running and avoid disruptive events such as a process upset or an unplanned plant shutdown. To avert such situations some maintenance managers opt to hedge their requirements for parts safety stocks. As a result, many organizations incur higher inventory carrying costs with excess materials spares stocks on hand. Artificial intelligence and Machine learning can help to reduce costs and improve stock management by anticipating, through algorithms, the parts demand based on historical material usage data and installed equipment failure rates.

Organizations will likely see value in the establishment of a data center that works to optimize analysis models and drive further data capture. A data center organization is a key component of any successful digital transformation initiative as it enables collaboration between business and data analysis experts around the use of Artificial Intelligence and Machine Learning.  As such capabilities require specialized skills, it may be necessary to augment the data center with external resources who possess specific niche subject matter expertise. However, in this operations context, data analysis skills that may be sourced externally need to be quickly integrated into the company to secure competitive position and preserve data security.

The main goal of data analysis is for derived insights to be central to operational decision making. It is therefore essential to consider the following:

  • Make sure relevant and reliable information is available at the right time: choose and identify key data sources with added value, define the relevant analysis frequency and how the final information will be communicated to the end user.
  • Build new predictive models: predictive models should help the industry rethink a new way of operating. Existing processes and the management of data across its life cycle will be impacted, as digital will bring a large amount of data in real time from multiple sources
  • Prepare and anticipate the transition from external to internal data analysis knowledge: guarantee autonomy and appropriation of new tools and methods (data models will become a competitive differentiator for companies and should be considered and treated as organizational intellectual property (IP).
  • Ensure that a culture of operational excellence is pervasive: drive further optimization of systems and processes to increase the value of data

Understanding and developing the above points are excellent starting points for any organization seeking to make sense of the significant amount of data generated and leverage that data for analysis that supports value-adding insights. However; moving to a more predictive mindset and model will require an understanding of both the human and IT impacts, as well as a strategy for managing digitalization – both of which will be touched upon in the Part 2 and Part 3 of this series.