che_november-2024 - 36

Applied Technologies
AI-Powered Search Improves Knowledge
Transfer at Agricultural Chemicals Site
An artificial intelligence (AI)-based search tool allowed workers to quickly mine
historical data to improve operations and respond to emerging problems
Andreas Eschbach
eschbach GmbH
t the Bayer Crop Science
facility in Muttenz, Switzerland,
managers and workers
wanted to improve communication
during shift handovers
and enable more efficient knowledge
transfer. The site had already digitized
its shift handover notes, giving
personnel a vast repository of historical
data, but its next challenge was
how to locate relevant information
quickly on the shop floor. A smart
search system powered by artificial
intelligence
(AI) has helped them
mine these data to drive operational
improvements and respond quickly
to emerging issues.
Inefficient knowledge transfer
" Knowledge transfer is a big issue
in shift operations, " says Matthias
Hesskamp, head of site operations
and excellence at Bayer Muttenz.
" Shifts are producing and generating
data 24/7, but the Day Operations
staff, who are responsible for
problem solving, are only onsite
working 40 hours per week, " he
explains. " That team faces the challenge
of processing data that are
generated constantly every day. "
To enable better communication
across shifts and teams, Bayer
Muttenz implemented a digital shift
handover system, known as Shiftconnector
by eschbach GmbH
(Bad Säckingen, Germany; www.
eschbach.com), about ten years
ago. The software now acts as a
centralized database and communication
platform, capturing shift
notes and other critical plant data
in one location (Figure 1). This improves
information flow and transparency,
since employees know
where to find updated information
from recent shifts that they need
36
A
Eschbach
FIGURE 1. Natural language processing (NLP) allows employees to use plain language, rather than keywords,
to query the data contained in the shift handover system
to do their jobs. Over time, Shiftconnector
has become a valuable
repository of historical knowledge.
Bayer Crop Science process
specialist Sven Kränkel explains,
" Shiftconnector is mostly kept up
to date by the shift teams working
on site. They enter data on how the
plant is operated, any disruptions
or incidents, what ran well, and
what optimizations were implemented.
The Day Operations team
needs exactly these pieces of information
to assess whether the production
went to plan. Did we have
any disruptions? Were they recurring
ones? This allows us to evaluate
the
production
process and
plan ahead for the coming days,
weeks and months. "
However, finding the information
at the time it was needed proved
to be a challenge - especially if
entries were incomplete or searchers
did not know the exact correct
search term needed to uncover the
information. Using standard keyword-based
search tools was inefficient
and did not always return the
most relevant results.
Shift collaboration system
Eschbach worked with Bayer Crop
Science in Muttenz to develop a
customized Smart Search tool with
AI that could be used inside Shiftconnector.
Natural language processing
(NLP) allows employees to
query the system in plain language,
without worrying about exact keywords.
An example query could
be, " has this product output ever
turned brown before? " . NLP allows
the system to analyze unstructured
data like manual shift notes
to understand their contents and
find relevant results, while machine
learning (ML) enables it to look for
patterns in vast amounts of data.
To make the system usable, the
AI had to be trained on domain- and
site-specific language, including
technical terms and abbreviations.
Eschbach worked with Bayer Crop
Science and leading AI researchers
at the University of Göttingen to
adapt an off-the-shelf AI search tool
for their needs. It took two years of
development, prototyping and beta
testing, which included user groups,
workshops and onsite investigations
to gather insights into users'
CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM
NOVEMBER 2024

http://www.eschbach.com http://WWW.CHEMENGONLINE.COM

che_november-2024

Table of Contents for the Digital Edition of che_november-2024