Paper Technology International 2024 - Journal - Page 20
PAPERTECHNOLOGYINTERNATIONAL
Detecting deviations in process harmony
At the core of this novel way to optimize the papermaking
process is the concept of process harmony. In practice, Harmonizer
monitors multiple key parameters and variables in the process to
ensure that the process is stable – in other words, in harmony.
Deviations in process harmony indicate that there are emerging
issues in the process and that the risk for production disturbances
is increasing. Early detection enables fast reaction and corrective
measures.
The unique view into the papermaking process is made
possible with harmony modeling. A harmony model estimates if a
particular process parameter, a tag, is in harmony with other tags at
the plant. In harmony modeling, the value of a dependent variable
is estimated based on explanatory variables. Here’s a simple
example: a harmony model monitoring the out昀氀ow of a 昀椀lter. In this
case, the explanatory variables for the out昀氀ow measure would be
the in昀氀ow, pressure measurements, and other data tags that are
closely related; these connections have been automatically found
by the machine learning model, without human interaction. If the
昀椀lter started to clog for any reason, harmony would decrease as the
out昀氀ow would no longer match the other variables. The harmony
deviation highlights that there is something wrong with the 昀椀lter and
allows operators to investigate the issue in time, even if it does not
pinpoint the exact issue.
There are thousands of tags at a single mill site. To have
enough coverage to monitor the papermaking process, the model
building is highly automated. Also, the retraining of the harmony
models is automated.
Automated modeling ensures that the harmony analysis is
easy to deploy at a mill site and that the models are always up to
date. Users don’t need to spend time building or training the model,
but they can focus on utilizing the provided information and insights
for continuous development in their operations.
From a data perspective, all the interconnected
variables in a paper machine could look like this [Figure 1].
The lines represent how different tags are connected to
each other, and the size of the dots represents how
strongly the speci昀椀c tags relate to each other. The
colors are based on different clusters of tags.
The picture illustrates the power of algorithms in
昀椀nding meaningful connections and correlations
in the data. As an example: the blue dots
represent the heating system. Different tags
related directly to the heating system are
naturally tightly connected to each other, but
as heat is a central part of a paper mill, the
blue tags correlate with multiple other tags in
the picture as well.
Harmony modeling can be described
as advanced monitoring. Many automated
models explain the current harmony in the
papermaking system by estimating each tag
value based on other tags. As a result, the
harmony level is not a 昀椀xed statistical minimum
or maximum value of a variable but a dynamic
calculation of its optimum level derived from other
measurements.
Figure 1: A paper machine as seen by algorithms.
20
All disharmonies are not equally critical to process stability
and runnability, but it is critically important to understand which are.
Harmonizer helps operators and engineers at the mill effectively
manage the process’s 昀椀ndings by drawing their attention to the most
important issues for process stability and supporting the ef昀椀cient
implementation of needed process changes.
Chemistry either makes it or breaks it
A stable papermaking system cannot be built by
understanding only the mechanical part of the process or focusing
on paper physics through optimized re昀椀ning. Chemistry plays a
crucial role in the papermaking system, both for production ef昀椀ciency
and end-product quality. The ability to better manage and optimize,
e.g. wet end chemistry performance, provides vast opportunities,
e.g., through reduced machine downtime and rejected reels,
decreased water consumption, energy, raw material, or optimized
chemical usage.
End-product quality
Chemistry greatly impacts the quality parameters of
the 昀椀nal paper or board: strength, hydrophobicity,
formation, smoothness, bulk, ash content...
Energy consumption
Chemical stability in昀氀uences retention and dewatering
performance, directly impacting sheet moisture after the
press section. Focusing on wet end applications helps
create conditions for optimum dewatering and thus
reduces energy consumption in the dryer section.
Deposition
Chemical stability also plays a role in the
physicochemical colloidal state of the process through,
