Sustainability,
Journal Year:
2024,
Volume and Issue:
16(21), P. 9555 - 9555
Published: Nov. 2, 2024
For
decades,
fossil
fuels
have
been
the
backbone
of
reliable
energy
systems,
offering
unmatched
density
and
flexibility.
However,
as
world
shifts
toward
renewable
energy,
overcoming
limitations
intermittent
power
sources
requires
a
bold
reimagining
storage
integration.
Power-to-X
(PtX)
technologies,
which
convert
excess
electricity
into
storable
carriers,
offer
promising
solution
for
long-term
sector
coupling.
Recent
advancements
in
machine
learning
(ML)
revolutionized
PtX
systems
by
enhancing
efficiency,
scalability,
sustainability.
This
review
provides
detailed
analysis
how
ML
techniques,
such
deep
reinforcement
learning,
data-driven
optimization,
predictive
diagnostics,
are
driving
innovation
Power-to-Gas
(PtG),
Power-to-Liquid
(PtL),
Power-to-Heat
(PtH)
systems.
example,
has
improved
real-time
decision-making
PtG
reducing
operational
costs
improving
grid
stability.
Additionally,
diagnostics
powered
increased
system
reliability
identifying
early
failures
critical
components
proton
exchange
membrane
fuel
cells
(PEMFCs).
Despite
these
advancements,
challenges
data
quality,
processing,
scalability
remain,
presenting
future
research
opportunities.
These
to
decarbonizing
hard-to-electrify
sectors,
heavy
industry,
transportation,
aviation,
aligning
with
global
sustainability
goals.
Data-Centric Engineering,
Journal Year:
2025,
Volume and Issue:
6
Published: Jan. 1, 2025
Abstract
This
article
establishes
a
data-driven
modeling
framework
for
lean
hydrogen
(
$
{\mathrm{H}}_2
)-air
reaction
rates
the
Large
Eddy
Simulation
(LES)
of
turbulent
reactive
flows.
is
particularly
challenging
since
molecules
diffuse
much
faster
than
heat,
leading
to
large
variations
in
burning
rates,
thermodiffusive
instabilities
at
subfilter
scale,
and
complex
turbulence-chemistry
interactions.
Our
approach
leverages
Convolutional
Neural
Network
(CNN),
trained
approximate
filtered
from
emulated
LES
data.
First,
five
different
premixed
-air
flame
Direct
Numerical
Simulations
(DNSs)
are
computed
each
with
unique
global
equivalence
ratio.
Second,
DNS
snapshots
downsampled
emulate
Third,
CNN
as
function
scalar
quantities:
progress
variable,
local
ratio,
thickening
due
filtering.
Finally,
performances
model
assessed
on
test
solutions
never
seen
during
training.
The
retrieves
very
high
accuracy.
It
also
tested
two
filter
downsampling
parameters
ratios
between
those
used
For
these
interpolation
cases,
approximates
low
error
even
though
cases
were
not
included
training
dataset.
priori
study
shows
that
proposed
machine
learning
able
address
challenge
rates.
paves
way
new
paradigm
simulation
carbon-free
combustion
systems.
