COVER FEATURE BATTERY HEALTH
& SAFETY
designed for virtually unrolling
papyrus scrolls but can be used to
nd out exactly what happens in
compact densely wound batteries
– as a result the characteristic
problems associated with wound
batteries are now open to
investigation.
According to Tengattini, the
research has thrown up some
interesting ndings.
“We found that the inner windings
exhibited completely different
electrochemical activity than the outer
windings. In addition, the upper and
lower parts of the battery behaved
very differently.
“The neutron data also showed
areas where a lack of electrolyte
developed, and that can severely
limit the functioning of the respective
electrode section. It could also be
shown that the anode is not equally
well loaded and unloaded with lithium
everywhere.”
According to Tengattini the
research not only showed pockets
within the battery where acceleration
and depletion occurred much faster
but also parts of the battery that were
getting completely closed off.
“To date, we’ve only conducted
this research on a small number of
batteries. While we have observed
some patterns more detailed
research will likely throw up more
reasons for why batteries fail. There
are a number of elements that we
still need to better understand.
“Despite that this research will
for improving the design of wound
batteries.”
Tengattini concluded, “As with all
nite resources, we can expect the
‘electric revolution’ to lead to greater
less resource. So, in order to meet
this, we need to better understand
what is happening at the heart of
these cells and while the technique
we have developed might lead to
revisions in terms of how batteries
are designed in the future it is too
early to say. Despite that it is an
important step forward.”
Machine Learning techniques
While that particular team of
scientists have been investigating
the physical structure of the battery,
scientists from Cambridge and
Newcastle universities, have designed
a machine learning method that they
claim is able to predict battery health.
According to the scientists
involved the method provides 10x
higher accuracy than current industry
standards.
The technique works by sending
electrical pulses into the battery and
measuring the response. They are
then measured and processed by a
the battery’s health and lifespan. It’s
a non-invasive technique and is a
simple add-on to any existing battery
system.
“Safety and reliability are the
most important design criteria as
we develop batteries that can pack
a lot of energy in a small space,”
explained Dr Alpha Lee from
Cambridge’s Cavendish Laboratory,
who co-led the research. “By
improving the software that monitors
charging and discharging, and using
data-driven software to control the
charging process, I believe we can
power a big improvement in battery
performance.”
The researchers have performed
over 20,000 experimental
measurements to train the model, the
largest dataset of its kind.
Importantly, the model learns how
to distinguish important signals from
irrelevant noise.
The researchers have been able
to show that the machine learning
model can be interpreted to give
hints about the physical mechanism
of degradation. The model can
inform which electrical signals are
most correlated with aging, which in
turn allows them to design speci c
experiments to probe why and how
batteries degrade.
The researchers are now using
their machine learning platform to
understand degradation in different
battery chemistries. They are also
developing optimal battery charging
protocols, powering by machine
learning, to enable fast charging and
minimise degradation.
Predicting the state of health
and the remaining useful lifespan
of lithium-ion batteries is critical as
new technologies come online and
demand better battery technologies.
According to the scientists
involved in these various projects
this work will help to aid in the
development of safer and more
reliable batteries for a host of new
devices – from electric vehicles to all
forms of consumer electronics.
12 28 April 2020 www.newelectronics.co.uk
© T.Arlt, I. MankeHZB, R. ZiescheUCL
Above: The x-ray
tomography shows
ruptures (black)
in the regions of
electrical contacts
(white)
Below: Neutrons
can detect dry
regions (yellow
arrow) where the
elecrolyte is lacking.
The blue arrow
shows areas with
a defi ciency of
Lithium.
www
help us to develop speci c strategies
demand from lithium-ion batteries, for
machine learning algorithm to predict
/www.newelectronics.co.uk