Page 148 - Intelligent Digital Oil And Gas Fields

Components of Artificial Intelligence and Data Analytics 111

2 4
MODEL DIAGNOSTICS
DATA EXPLORATION
VALIDATION
MISSING DATA
QUANTIFICATION
OUTLIERS
INTERPRETATION
DATA RELATIONS VISUALIZATION
1 3 5
DATABASE
ACQUISITION MACHINE LEARNING
INTEGRATION MODEL OPTIMIZATION
AGGREGATION
Fig. 4.3 Proposed modular advanced data analytics workflow for the E&P industry,
where components of individual analytics domains, from descriptive to prescriptive,
merge into a collaborative synergy.

The workflow continues with Modules 3 and 4, which combine selec-
tion and building the predictive analytics (e.g., ML) model (for ML model
selection, see Section 4.2) as well as validation, quantification, interpreta-
tion, and visualization of results. The workflow ends with Module 5 and
the prescriptive analytics phase, where the results of the predictive model
provide an input for a nonlinear optimization, where certain KPIs can be
defined as minimization [e.g., cycle time or nonproductive time (NPT)]
or maximization (e.g., production, rate of penetration (ROP)) problems
via a suitable objective/cost function [e.g., in sparse equations and least
square (LSQR) form]. The benefits of deploying advanced data analytics
workflow in the modular form are as follows:
• A project can grow in functionality by adding project files for tasks.
• Intellectual property (IP) can be modularized within individual project
files, which makes collaboration easier.
• Modularizing promotes functionality reuse, unit tests, easier
documentation, etc.
4.1.3 Big Data in E&P: Concepts and Platforms

E&P operations have traditionally generated large volumes of data; how-
ever, with the advent of “smart operations” and DOF projects, the E&P
industry is now producing extreme volumes, at exponentially higher rates
than ever before. Today’s operations generate terabytes and petabytes of
data, at extremely large volumes, speeds, and acquisition frequencies from
multiple sources and domains, such as geophysical, geological, engineering,
production, surveillance, maintenance, etc. The E&P industry is quite liter-
ally experiencing data and information overload; it needs a focus and