Page 187 - Intelligent Digital Oil And Gas Fields

Components of Artificial Intelligence and Data Analytics 143

satisfactory operation or observed malfunction. This correlation and knowl-
edge enables a shift toward proactive ESP monitoring to predict and identify
potential problems long before they occur thereby reducing intervention
costs and optimizing production.

4.3.4 Real-Time Analytics on Streaming Data

This chapter concludes by describing a few trends and applications in the
area of data analytics that have the potential to transform E&P in the era
of Big Data and high-performance cloud computing. Traditionally, opera-
tors have collected massive amounts of data from equipment, fields and
assets, and daily operations. Mostly the data are stored and archived, then
processed and analyzed as required. Such traditional workflows result
in deterministic analytics and passive surveillance. However, SAS (2015),
aleadinganalyticsandsoftwarecompany,believesthatthreetechnologieswill
radically overhaul operational capabilities in the oil and gas sector: the IoT,
event streaming processing, and prescriptive analytics. Synergistically, these
technologies have the potential to rapidly deliver key operational insights
and make analytics tools available and impactful “at the point and at the time
of decision” by embedding analytics into decision-making and operational
processes. The key to implementing ESP technologies lies in enabling con-
textual and situational analytics “on the fly”—in ultra-speed, ultra-low-
latency environments—which creates a technological inflection point for
the next-generation remote-control operations. However, we need to note:
at this evolutionary stage of fast data and streaming analytics, E&P operations
still do not benefit from “real real-time” analytics (i.e., seconds, milliseconds,
microseconds), but rather operate at “near real time,” approximately several
seconds and more—much more when recovering from infrastructure faults.
The concept of stream processing computation or stream computing con-
sists of assimilating data readings from the collections of software or hardware
sensors in stream form (i.e., as series of sequences or continuous queries),
analyzing the data “on the fly,” and producing actionable results, preferably
in real- or at the right time.
In current E&P practice, one area that benefits most from event stream-
ing processing and prescriptive analytics is drilling operations. Despite
the arguments that drilling analytics are not yet oriented toward automati-
cally maintaining the knowledge base through “near real time” updates,
Staveley and Thow (2010) have reported that enabling the results in
“near real time” is considered to be highly valuable because it provides