42
NuclearPlantJournal.com Nuclear Plant Journal, July-August 2016
Predicting
the Plant
Issues
By Mona Badie, GE Hitachi Nuclear
Energy.
Mona Badie
Mona Badie is Chief Information
Officer and Chief Digital Officer for GE
Hitachi Nuclear Energy. Badie joined
GEH in August 2015
and is responsible
for all information
technology systems
and processes
including the
company’s Industrial
Internet efforts.
A 12-year GE
veteran, Badie
has held several
leadership roles
including Executive
IT Leader for Leasing
Originations and
Executive IT Leader
for Enterprise
Analytics at GE
Capital. Before
joining GE she
held leadership positions at Fitch Risk
and Polaroid and was the Robotics
Department Chair at ITT Technical
Institute. She started her IT career as a
programmer at Polaroid.
Mona earned a Master of Science
Degree in Electrical Engineering from
Purdue University and a Bachelor in
Electrical Engineering from Alexandria
University, Egypt.
An interview by Newal Agnihotri, Editor
of Nuclear Plant Journal, at the Nuclear
Energy Institute’s Nuclear Energy Assembly
in Miami, Florida on May 23, 2016.
1.
Are the GE Hitachi digital solutions
versatile across various reactor plants?
Yes. GE Hitachi Digital solutions are
not reactor specific. In fact, one of our
Exelon pilots is at a PWR plant, while the
other pilot is across plants with PWR and
BWR technologies.
2.
What are different applications of
Predix?
We have multiple applications and
pilots based on Predix.
The Asset Performance Management
suite allows plants to reduce unplanned
downtime and prevent lost power
production by connecting their assets,
alerting operators when anomalies
happen,
predicting
failures usingmachine
data and analytics,
and optimizing the
maintenance profile.
The
Resource
planning and In-
Processing
suite
provides a centralized
view
of
staffing
requirements
and
personnel availability.
This enables staffing
managers to optimize
work assignments.We
are also working on a
pilot to digitize In-
Processing to reduce
overhead costs.
Our co-innovation
pilot with Exelon is
focused on human
performance and the prediction and
prevention of negative events.
3.
Are the digital solutions being
implemented in different phases at
different plants?
We currently have two pilots with
Exelon; Watchtower and Lighthouse.
Both pilots are in phase 2.
4.
What’s the difference between
Watchtower and Lighthouse?
Watchtower is based on the Asset
Performance Management suite. It
focuses on asset/equipment reliability
and maintenance optimization.
Lighthouse is a co-innovation pilot.
It uses human performance data to predict
Plant Key Performance Indicators.
All nuclear operators collect a lot of
human performance data. Is all of this
data valuable? Can it be used to predict
negative events?
5.
Can you predict events at the plants
in advance, using current applications?
Yes! This is the premise of
“Lighthouse”, our Human Performance
co-innovation pilot with Exelon. The
question we asked was: “Can we utilize
historical data to predict the outcome of
performance indicators?”
We worked with Exelon and selected
6 KPI’s (Key Performance Indicators) to
test: Capacity Factor, Forced Loss Rate,
Reactivity Management, Unplanned
Manual and Automatic Scrams, Safety
System Performance and Event-free
Clock Resets. We were able to predict
negative events 3 to 6 months in advance
with high accuracy!
6.
How is the event prediction made?
We started with 800 data points
across multiple functions and were able
to narrow them down to 20 relevant
factors per KPI.
We created an analytical model and
trained it using five years of historical
data across 13 plants. We then ran new
data through the models and were able to
predict the negative events in advance.
7.
What’s the difference between Asset
Performance Management System and,
“Predix”?
Predix is an industrial, system-
agnostic platform. This is where the data
is stored, and applications and analytics
are performed. Asset Performance
Management (APM) is an application
suite, built on Predix. It collects data from
sensors, alerts operators when anomalies
happen, and uses analytics to predict
failures. Our Watchtower pilot with
Exelon is based on APM.
8.
Were the related sensors pre-existing?
Yes, the assets we monitored with
the Asset Performance Management pilot
were already equipped with sensors. We
monitored balance of the plant: Main
turbines and auxiliaries; Feedwater
systems and condensate; Feedwater
heaters, heater drains, extraction steam
