Page 150 - Intelligent Digital Oil And Gas Fields

Components of Artificial Intelligence and Data Analytics 113

Enhance safe and
Improve asset value
environmental-friendly
and returns
operations
Interpretation
Prediction
Improve production, Increase efficiency and
recovery and Advisory productivity across
facilities efficiency Scalability major business units
Collaboration
Optimize exploration,
drilling, and production Reduce operational
operations costs
Fig. 4.5 Main areas of interaction between the attributes of Big Data analytics and E&P
business segments, with the most potential to add value.

is to extract, load, and transform. The new paradigm is to collect and load
data into the Apache Hadoop open source database (Ghemawat et al.,
2003; Handy, 2015), which enables distributed processing of large data
sets on clusters and servers, without extensive transformation into a rela-
tional database model for further analysis.
• Velocity: this component relates to the understanding that the acquired
data are no longer data at rest (or static) and adopting new methods
for data in motion (e.g., streaming or fast data) to analyze data in real
time. Not all data received in real time need real-time analysis. However,
some (e.g., real-time alerts for operational efficiency and failure diagnos-
tics) need real-time adaptive analytics with stream computing and
support of massively parallel-processing databases (Brul e, 2009) and
low-latency data-flow architecture (Brul e, 2013).
• Variety: Big Data consist of structured and unstructured data. While struc-
tured data are generally in digital form, acquired by sensors (e.g.,
temperature,pressure,fluidflow),theunstructured(noformat)datacome
in the form of text files, well files, field development reports, drilling
records, etc. (see Table 4.1) and requires specific types of text analytics
(down into Boolean operands) to extract information at large scales.
• Veracity: this component relates to the accurateness and correctness of
data. In circumstances of the first 3 Vs, confusion can arise because of
incomplete (and sometimes obscure) definitions of how true and trust-
worthy are the data?
• Virtual (data): this component of Big Data enables the E&P industry to
generate abstracted and integrated information in real time, from dispa-
rate sources, and send it to multiple applications and users. The virtual
data centers/servers are easier to build and consume (than traditional data
stores), and require much less effort to maintain.