Materials,
Journal Year:
2024,
Volume and Issue:
17(24), P. 6280 - 6280
Published: Dec. 22, 2024
Industrial
and
construction
wastes
make
up
about
half
of
all
world
wastes.
In
order
to
reduce
their
negative
impact
on
the
environment,
it
is
possible
use
part
them
for
concrete
production.
Using
experimental–statistical
modeling
techniques,
combined
effect
brick
powder,
recycling
sand,
alkaline
activator
fresh
hardened
properties
self-compacting
production
textile-reinforced
was
investigated.
Experimental
data
flowability,
passing
ability,
spreading
speed,
segregation
resistance,
air
content,
density
mixtures
were
obtained.
The
standard
ability
tests
modified
using
a
textile
mesh
maximize
approximation
real
conditions
To
determine
dynamics
strength
development,
compression
flexural
at
ages
1,
3,
7,
28
days
splitting
tensile
conducted.
preparation
technology
investigated
depending
composition
presented.
resulting
mathematical
models
allow
optimization
compositions
partial
replacement
slag
cement
with
powder
(up
30%),
natural
sand
recycled
100%)
addition
an
in
range
0.5–1%
content.
This
allows
us
obtain
sustainable,
alkali-activated
high-strength
concrete,
which
significantly
reduces
environment
promotes
development
circular
economy
industry.
Buildings,
Journal Year:
2025,
Volume and Issue:
15(4), P. 593 - 593
Published: Feb. 14, 2025
Sandwich
panels,
consisting
of
two
concrete
wythes
that
encase
an
insulating
core,
are
designed
to
improve
energy
efficiency
and
reduce
the
weight
construction
applications.
This
research
examines
thermal
flexural
properties
a
novel
sandwich
panel
incorporates
ultra-high-performance
fiber-reinforced
(UHPFRC)
cellular
lightweight
(CLC)
as
its
core
material.
Seven
specimens
were
tested
for
their
thermo-flexural
performance
using
four-point
bending
tests.
The
experimental
parameters
included
variations
in
UHPFRC
thickness
(20
mm
30
mm)
different
shear
connector
types
(shear
keys,
steel
bars,
post-tension
bars).
study
also
assessed
effects
adding
mesh
reinforcement
layer
evaluated
box
sections
without
CLC
core.
analysis
concentrated
on
several
critical
factors,
such
initial,
ultimate,
serviceability
loads,
load–deflection
relationships,
load–end
slip,
load–strain
composite
action
ratios,
crack
patterns,
failure
modes.
transient
plane
source
technique.
results
demonstrated
panels
bars
connectors
achieved
performance,
most
favorable
ratios
reached
68.8%.
Conversely,
section
exhibited
brittle
mode
when
compared
other
tested.
To
effectively
evaluate
mechanical
properties,
it
is
important
design
have
adequate
load-bearing
capacity
while
maintaining
low
conductivity.
introduced
thermo-mechanical
coefficient
both
panels.
findings
indicated
with
highest
those
had
lowest
performance.
Polymer Composites,
Journal Year:
2025,
Volume and Issue:
unknown
Published: Feb. 19, 2025
Abstract
Fiber‐reinforced
polymers
(FRPs)
are
widely
recognized
as
ideal
materials
for
transport
structures
due
to
their
customizable
properties,
high
strength,
and
stiffness
combined
with
low
density.
These
exhibit
significant
resistance
atmospheric
conditions
but
susceptible
fatigue
loading.
Unlike
conventional
metals,
which
possess
an
endurance
limit,
FRPs
prone
failure
under
any
external
load
when
subjected
a
substantial
number
of
cycles.
This
makes
the
estimation
residual
strength
critical
aspect
composite
engineering.
study
evaluates
efficacy
various
machine
learning
(ML)
regression
models,
integrated
informatics,
in
predicting
post‐fatigue
carbon‐
glass‐based
(CFRPs
GFRPs).
A
total
10
features
that
closely
affects
behavior
were
used
train
ML
models;
five
from
manufacturing
aspects,
four
testing
parameters,
one
representing
properties.
The
tested
models
linear,
non‐linear,
decision
tree,
ensemble,
support
vector,
artificial
neural
network
(ANN)
approaches
identify
best
fit
dataset.
R‐squared
(R
2
),
Mean
Absolute
Error
(MAE),
Median
(MedAE)
Root
Square
(RMSE)
evaluation
metrics
assess
model
performance.
findings
indicate
available
numerical
data
is
sufficient
initiate
training
develop
robust
prediction,
though
scope
improvement
remains
expansion
Among
Multi‐Layer
Perceptron
(MLP),
ANN‐based
regressor
two
hidden
layers
comprising
30
20
neurons,
achieved
performance,
R
values
0.88
on
validation
set
0.95
test
RMSE
72.42.
Additionally,
tree
(DT)
AdaBoost
regressors
recorded
MedAE
zero
data,
suggesting
at
least
half
predictions
accurate.
boosted
DT
also
demonstrated
lowest
dataset,
value
2.13.
Highlights
Data
compiled
literatures
contains
outliers
degrades
model.
Feature
importance
accuracy
models.
ANN
takes
highest
time
presents
fit.
GridSearchCV
improves
prediction.
Models
presenting
negative
can
be
improved
by
varying
parameters.
