Hydrocarbon Engineering - July 2020

23 / 76

July 2020 HYDROCARBON
ENGINEERING 21
traditional techniques, such as linear regression and various
‘rules of thumb’. Refiners tend to clean the most fouled
bundles rather than the bundle with the greatest economic
value. CPM uses Petro-SIM and the most advanced prediction
techniques to project future fouling factors. It uses rigorous
heat and material balances to calculate the right bundle to
clean based on economic outcomes. The CPM predictions
and simulations empower the refinery’s asset management
teams to determine the right bundles to clean at the right
time (Figure 1).
CPM provides refinery operators with the capability to
predict future fouling in preheat trains several months in
advance. This allows them to reduce energy consumption
and avoid throughput-limiting conditions before they occur.
In addition, the solution provides insight on how to
optimise heat exchanger cleaning schedules to maximise
the return on investment.
Digital
Until recently, operators used process simulation models for
activities such as basic design, unit optimisation, crude
selection, and operator training. The simulations existed in
isolation of plant data models that the operator hosted in
information management systems. Users operating off the
plant’s information management systems have historically
been unable to directly access the value of the site’s process
simulation tools, and vice versa.
However, enhanced connectivity has rectified this by
permitting increased contextualisation of data and
multi-purpose use of single models. With added context and
relationship, more plant process data measurements can
constitute information. Ever-increasing context,
connectedness, and patterns can generate more insight for a
greater number of consumers.
With this IT foundation and by expanding upon data
management capabilities, process simulation tools have now
become a source for sustained value delivery. They provide the
answer to operating companies seeking digitalisation solutions.
Next generation process simulation tools can be
operationalised into digital twins with real-time production
data. By using this data from sites’ distributed control systems,
historian, and laboratory information systems, the digital twin
enables quick and informed decisions. The digital twin delivers
real-time, high-fidelity, virtual representations of hydrocarbon
molecule transformation and associated plant operating
conditions. All outputs of the digital twin are written into fully
historized relational databases for data mining. Furthermore,
the digital twin automatically validates mass balances and
reconciles process data, closing the value gap between
production plan and actual. Using the simulator in this way
enables real-time asset performance monitoring, surveillance,
supply chain optimisation, and other advanced applications
and services (Table 1).
Any changes to a data point or stream triggers automatic
notification to the simulation model. All model outputs
automatically write back, in real-time, thereby enhancing the
quality and richness of historian data. This includes
comparison of measured vs simulation model vs linear
programming model outputs to help track when models and
actual plant performance diverge.
Case study 3
Implementing a refinery digital twin is a key project that is part
of a European integrated oil company’s digital transformation
efforts. The digital twin outcomes will involve process
improvements that maximise production while optimising
energy consumption.
The refinery digital twin project integrates KBC’s
Petro-SIM process simulation software and OSIsoft PI-AF. By
leveraging existing simulation models with the latest ability
to connect with PI-AF, the operator will be able to closely
monitor the units.
As a result of continuous monitoring, the operator was able
to update the models in an agile manner. It achieved its goal of
generating more reliable production plans, allowing the
operator to identify pattern changes in the behaviour of the
main units, e.g. operational variations or catalyst cycle changes.
It is then possible to optimise the main process unit’s
performance by alerting when there are deviations between
actual data, unit simulator data, and planning data.
Entering a digital collaboration is difficult yet exciting. It
requires a company to recognise that it is lacking in certain
skills that partners can bring to the table. There must be value
to both sides. The obvious question then, if there is value, is
why not build that knowledge and skill in house?
There are times when acquisition is the right answer, while
other times a partnership is preferable. Ultimately, it is a
business decision about resources, priorities, and cost. It is
important to recognise that a company does not just buy
technology, but also the culture. Picking business partners
based on their capabilities and shared digitalisation values and
philosophy is the best method for success.
To further enhance collaboration success, the digital
process simulator should be built ‘open’ and with connectable
tools. It creates a modular digital twin that allows those who
want to progress further access to a full technology stack, each
built by experts that are passionate about their area of study.
User experience is also important. Understanding how they
want to interact with the technology stack plays a
development role. Being adaptable to all the stakeholders is
key to success.
Reference
1. ‘Achieving fullstream digitalization with continuous process
management’, Baker Hughes whitepaper, (2018),
com/document/white-paper-accenture-and-kbc-achieving-fullstream-
digitalization-continuous-process-management
Table 1. Always aligned with and driving the
business data model
Traditional simulator Digital twin
An accurate
representation of a
specific operating case
An accurate representation of
the asset over its full range of
operation, all of the time
Static provision of a
snapshot in time
Captures the full history and future
of the assset
Built on an ad-hoc basis
to answer a question
Automated, regular model runs.
Built-in to business workflows
Owned and used by
isolated groups on an
ad-hoc basis
Centralised single version of the
truth, everyone uses, delivers
outputs directly to the business,
strong governance systems