Page 211 - Intelligent Digital Oil And Gas Fields

Workflow Automation and Intelligent Control 165

of “cognition” programmed into the workflow and the capacity for reason-
ing beyond the physical laws that have been programmed into the workflow.
Scientists have introduced a series of soft computing techniques (neural net-
works, fuzzy logic, pattern recognition, etc.) to help workflows improve
efficiency and include a level of smartness in the process. However, these
techniques require a significant amount of human development and
debugging to implement (as discussed in Chapter 4). Soft-computing tech-
niques are difficult to initially implement in automated workflows, but once
implemented they enable the workflows to move from merely automated to
smart and then truly advanced by doing human functions such as “learn” and
acquire “expert” knowledge. Al-Abbasi et al. (2013) and Al-Jasmi et al.
(2013a) suggest that the following pillars should be considered during the
design of smart workflows:
Knowledge capture. The benefit of knowledge capture is to standardize
processes and assure that recurring tasks are done consistently, without
fail, and include monthly best practices capable of improving asset per-
formance. Knowledge capture is also important to transfer knowledge
from subject matter experts (SMEs) to less experienced engineers. Skin
factor increasing with time in oil wells or preliminary values of pump
wear factor after 2years of production are typical examples of knowledge
captures. Workflows can be designed to show smart tips that help less
experienced personnel understand why data relates to a physical law
or that data is out of acceptable range.
Continuous improvement or “Kaizen.” A smart workflow will continue to
improve and learn if it can recognize patterns in the data and knowledge
that it captures. Artificial intelligence systems can recognize patterns in
operational variables and predict behaviors, which result in continuous
process improvements. Today, technology can generate a self-trained
neural network. However, when new events happen, the neural net-
work must be tuned; therefore, there is no complete set of soft computer
with a full awareness process.
Multidisciplinary collaboration. Smart workflows should be designed for a
high degree of collaboration among disciplines along the E&P value
chain, such as subsurface, reservoir, production, operations, drilling,
and surface facilities. When working in the system, each discipline should
contribute to the main goal and help overcome the complexity of oper-
ational problems. Smart workflows use technologies that can support
intensive, high-traffic, cross-discipline data flows, information sharing,
and knowledge integration.