I SSUE 5

20
23
E N GL I S H

VISIT US AT: TECHNICAL FIBERS, TECHNICAL D 3339 F

TEXTILETECHNOLOGY.NET TEXTILES, NONWOVENS

IN THIS
ISSUE

Fibers News
BASF / Asahi Kasei
Lenzing
Quantum
Materials
SGL Carbon
Spiber
Toray
Hemp
composting
ITMA review:
technical fiber
production

Technical Textiles
ITMA review:
weaving
Geosynthetics
market
Traces for
E-textiles
Nonwovens
market
41 TECHNICAL TEXTILES 5 | 2023
NONWOVENSTRENDS

Material uniformity assessment

technique for the nonwovens and
automotive industries

A novel real time evaluation of the uniformity of high-loft nonwovens

using patented in-line measurement technology is introduced. Autins
Alaa Memari Group, a specialist in thermal and acoustic materials for the automotive
Autins Group Plc.,
Rugby/UK
industry, utilizes data acquired by a patented Hammer-IMS industrial
basis-weight scanner and a modified uniformity index algorithm, origi-
Tom Redant
Hammer-IMS nv,
nally developed for full-area camera images, to provide a standardized
Herk-de-Stad/Belgium approach for assessing nonwovens uniformity. The use of this technique
was tested at Autins facilities in Tamworth/UK which is currently the
only nonwovens supplier in the automotive industry to implement it
and offer such a high level of quality control. The importance of non­
wovens uniformity in automotive applications, the manufacturing chal-
lenges, and the potential for standardization in the industry is shown.

The automotive industry is a well-regulated domain and de- property effectively. Currently the most used parameter to
mands high-quality and consistency from the materials for its describe nonwovens uniformity is the average areal density,
various applications, with a particular focus on improving acous- typically measured in g /m2.
tic and thermal performance. Nonwovens are commonly used in
automotive components for thermal and NVH (Noise, Vibration, Whilst meltblown, spunbond, carded, or hybrid materials are
and Harshness) purposes and are supplied by various tiers of preferred choices for automotive trim, thermal and NVH ap-
manufacturers. The nonwovens manufacturing process varies plications, they often face challenges in achieving a high level
depending on the specific application and the type of fibers of uniformity, particularly when using dated equipment, high-
employed. In the automotive sector, staple fiber dry-laid needle diameter fibers, or low material grammages. Although techni-
punched or thermo-bonded nonwovens as well as continuous ques like MicroPunch by Dilo and ProDyn by Andritz have been
filament spunbond or meltblown nonwovens are widely used. developed to improve uniformity in carded nonwovens, there is
currently no industrywide recognized standard for expressing
The success of nonwovens in providing acoustic and ther- the uniformity of nonwovens after production, leaving auto­
mal insulation hinges on their exceptionally high porosity, with motive supply chain organizations without clear quantitative
some high-loft materials achieving porosities exceeding 99%. methods for specifying material uniformity requirements.
In these materials, where 99% of the volume is air, only 1% is
occupied by the fibrous matter. Therefore, it is crucial to ensure Here, a solution is presented that combines mathematical
an even distribution. The insulation properties of such ma­terials meth­ods with technology to measure and profile areal density,
are highly sensitive to any irregularities or clustering in fiber as provided by the Hammer-IMS scanner, offering potential for
distribution. Since achieving perfect uniformity in nonwovens standardizing nonwovens uniformity assessments in the auto-
is inherently challenging due to the random distribution of motive industry.
fibers, a method for quantifying material uniformity is essential
for organizations in the industry’s supply chain to enable the
design, specification, selection, and communication of this

Next page
42 TECHNICAL TEXTILES 5 | 2023
NONWOVENSTRENDS

Production setting FIG. 2A

Concept drawing of the measuring principle. The 3 available

Autins employs a patented production technology 1 for manu- M-Ray heads create 3 so-called zig-zag-traces on the
facturing the high-performance NVH material known under their material. For the area lying along these traces, basis-weights
trademark Neptune. This material offers very good acoustic and are measured. Depending on the relative speed of the
thermal insulation at a significantly lower weight compared to material vs. the scanner’s scanning speed, the zig-zag cover-
similar materials in the automotive market. Hammer-IMS has age approaches a 100% surface coverage.
installed an industrial scanner in-line with the Neptune produc-
tion line. This scanner utilizes a time-of-flight principle 2 based
on high-frequency microwave signals, a technique branded as
Hammer-IMS’ (M-Rays), to measure the basis-weight distribu-
tion of the lofty polyester-polypropylene (PET/PP) nonwovens.
Unlike non-sustainable technologies like X-Rays or beta-radia- HMI measuring head

tion, this scanner offers a more eco-friendly approach.

FIG. 1A

Standalone Hammer-IMS scanner

scanner's
motion axis

ct g
n
re in
io
di nn
ru
web

FIG. 2B

Representation of
the matrix to which
the uniformity
region of interest
index algorithm is
applied.
running direction

FIG. 1B

Location of the scanner after inline installation on the

Neptune production line at Autins’ Solar Nonwovens plant in
Tamworth. The scanner is positioned in the production line
after the slitting stage and before the winding unit.

Another key advantage of using this scanning system is the

continuous measurement taken not only in machine direction,
but also in cross direction. This is thanks to the traversing move-
ment of the Marveloc-Curtain frame which carries the measu-
ring sensors (Fig. 1).

The scanner consists of 3 M-Ray measuring heads that gene-

rate traces of measurement data, combining them to provide
full-width basis-weight measurements in g/m2 (Fig. 2). The scan-
ner's resolution in the cross-machine direction is 2 mm, and it
operates using its control software, Connectivity 3.0, built on a
C++ runtime environment.
43 TECHNICAL TEXTILES 5 | 2023
NONWOVENSTRENDS

Uniformity index
FIG. 3

Dividing the image into quadrants is key in the algorithm.

Autins saw the opportunity to arrange the
scanner values in cartesian coordinates to
form an image-like map of material density.
This allows the implementation of statistical
image analy­sis methods such as the quad-
rant method using the uniformity index algo-
rithm. The quadrant method was originally
used in image processing and inspired by
ecological studies. The algorithm, as original-
ly presented by Amirnasr 3 for camera images,
is based on the quadrant method asses-
ses the variance between quadrants within
an image and how it changes as the number
of quadrants increases. Images with a high
level of color cluster­ ing result in increased
variance with more quadrants, yielding a lower uniformity index. This sliding-window dataset is generated by scan-data obtained
Conversely, images with a more homogeneous color distribu­tion by the Hammer-IMS scanner while it is scanning back and forth,
exhibit a lower increase in variance with additional quadrants, generating one additional line of data in the source matrix ‘dg’
resulting in a higher uniformity index. For example, Fig. 3 shows a for the algorithm. Each row of the source matrix ‘dg’ contains
gray­scale image of nonwovens taken at Autins laboratory. full-width scan data, equidistantly resampled in a 5 mm cross-
direction resolution. The typical width of 1,800 mm which Au-
On the left-hand side, the image is divided into 4 quadrants, tins uses corresponds to 360 values across the material width.
whereas on the right-hand side, the same image is divided into The dataset dg only keeps data from the most-recent 10 scans.
16 quadrants. The variance between quadrants in the image on Any older scan data is being removed from the dataset but
the left-hand side is lower than that of the right-hand side. This saved onto the Autins cloud for future analysis. The size of the
is due to the uneven, or clustered, distribution of fibers. The dg matrix is therefore 360 x 10 as shown in Fig. 2B. Since both
higher the number of quadrants, the higher the variance. The the product and the scanner are physically moving, each line in
change in variance represents the material uniformity. A per- the matrix does not correspond to a single position in the run-
fectly uniform image exhibits no change in quadrant variance, ning-direction of the material. Furthermore, the 3 zig-zag traces
yielding a uniformity index of 1 (or 100%). provide spare data since it has been generated by three discrete
M-Ray heads. This non-orthogonality and sparsity of the dataset
Hammer-IMS was able to integrate the algorithm with its control
software Connectivity 3.0 which assumes an OpenCV grayscale Next page
'Mat'-object as the data source.

The algorithm was adopted, but the histogram flattening was

left out to condition the input basis-weight data. Histogram flat- FIG. 4

tening can occur in many ways, and its effect on the remainder Uniformity Index modelled against the production line speed
of the algorithm is sometimes hard to predict, so we decided to
remove this from the algorithm. This means that the statistics Sample set Model 61.867* (1-e -1.3197x)
are not being altered in the same way as in the original unifor- 62
mity index algorithm. Instead, a single scale-factor K=200 was
applied to the index-of-dispersion I=K*VAR(X)/E(X). This value
61.5
has the effect of making the uniformity index more sensitive to
the typical variations in the considered nonwovens production
process for the case when histogram flattening is not present. 61

The result was a real-time map of material’s local density dis-

Uniformity index (%)

tribution and consequently, processed into a live assessment 60.5

of the uniformity during the production process of the Neptune
material. 60

The algorithm has been made open-source in both MATLAB 59.5

compatible code and C++ compatible code on a repository 6 to
encourage its standardization and contributions from other re-
59
searchers to be made. In the in-line application, the algorithm is
used with a sliding-window dataset generated by the scanner,
which captures data while scanning back and forth to create a 58.5
2 3 4 5 6 7
matrix that represents material density map.
Line speed (m/min)
44 TECHNICAL TEXTILES 5 | 2023
NONWOVENSTRENDS

Calendar varies based on the production line speed and is in the current
state-of-the-work not compensated for. Evaluation of the tech-
nique in the next section discusses this approximation.
2023 Trade Fairs and Conferences
December
Evaluation of the technique
13-14 AATCC Webinar on Haptics & Textiles, Virtual
→ www.aatcc.org/aatcc-events/haptics/
The technique was evaluated by testing its results of scanning
the same sample material at different line speeds (2.3, 4.3,
6.3 m/min) to assess the effect of dataset sparsity on the uni-
2024 formity index. The results show that variations in line speed im-
pact the uniformity index by no more than 5%. The correlation
January
between line speed and the uniformity index was modelled
9-12 heimtextil 2024, Place: Frankfurt/Germany using an exponential model (Fig. 4). As future work, the inclu-
→ www.heimtextil.de sion of a process-specific empirical model to mitigate this
11-14 Domotex 2022, Place: Hanover/Germany dependence is proposed. The sparsity of the dataset, due to
→ www.domotex.de the scanner's discrete M-Ray heads and non-orthogonal data,
20-22 Innatex 2024, Place: Hofheim-Wallau/Germany can be a challenge, particularly for faster production lines.
→ www.innatex.de

23-25 Munich Fabric Start, Place: Munich/Germany

→ www.munichfabricstart.com
Conclusion and future work
24-25 16. Bautextilien-Symposium, Place: Chemnitz/Germany A novel application of a uniformity-index calculation was intro-
→www.stfi.de/veranstaltungen/symposium-bautex duced to assess nonwovens uniformity in an in-line production
24-25 MENA Nonwovens Event 2024, Place: Dubei/UAE setting. The influence of the relative speeds of the production
→ https://edana.idloom.events/mena-nonwovens line and the basis weight scanner on the accuracy of the unifor-
31 Forum Funktionalisierung, Place: Hohenstein/Germany
mity index was discussed. Future work aims to develop a model
→ www.afbw.eu/aktuelles/veranstaltungen/details/forum- to compensate for variations in the uniformity index caused by
funktionalisierung-2024/ changing line speeds. Further research and experimentation
February would help to understand the effects of dataset sparsity on the
algorithm's application.
5-7 Texworld Evolution, Place: Paris/France
→ www.texworld-paris.fr
Ultimately, this work seeks to address the absence of standards
7 10th Congress Composite Simulation,
for assessing nonwovens uniformity in the nonwovens and auto-
Place: Augsburg/Germany
→ www.afbw.eu/aktuelles/veranstaltungen/details/10-fach- motive markets with this proposed solution and contributes
kongress-composite-simulation-2024/ open-source code to encourage collaborative standardizations
19-23 R+T 2024, Place: Stuttgart/Germany → www.rt-expo.com efforts.
22-24 20th International Istanbul Yarn Fair, Place: Istanbul/Turkey
Neptune, MicroPunch, ProDyn = trademarks
→ https://iplikfuari.com/en/?_ga=2.89351704.
260516422.1693916143-1850808752.1693916143

March

5-7 JEC World 2024, Place: Paris/France

→ www.jeccomposites.com

6-8 Composites 2024 International Conference, Place: Sevilla/

Spain → www.setcor.org/conferences/composites-2024

6-8 Intertextile Shanghai Home Textiles, Place: Shanghai/China

→ www.intertextilehome.com

13-14 Cellulose Fibres Conference, Place: Cologne/Germany References

→ www.cellulose-fibres.eu 1. Lee, H.J. et al: European patent “Waved meltblown fiber web
and preparation method therefor”, Patent No: EP2918718B1
18-22 International Week of Narrow and Smart Textiles,
Place: Dresden/Germany 2. Deferm, N.; Redant, T.; Dehaene, W.; Reynaert, P.: Patent
→ www.tu-dresden.de/mw/itm/narrow-2024 family “Sensor for non-destructive characterization of
objects”, Patent No.: WO2016198690A1
20 DITF-Innovationstag, Place: Denkendorf/Germany
3. Amirnasr, E.; Shim, E.; Yeom, B.Y.l.; Pourdeyhimi, B., Basis
→ www.ditf.de/innovationstag weight uniformity analysis in nonwovens, The Nonwovens
20-21 Performance Days, Place: Munich/Germany Institute, North Carolina State University, Raleigh, NC/USA.
→ www.performancedays.com Published online: September 5, 2013
4. MicroPunch, demonstrated at ITMA 2023, Milan/Italy,
April June 2023, www.nonwovens-industry.com/contents/view_
breaking-news/2023-06-23/dilo-details-success-at-itma/
23-26 Techtextil 2024, Place: Frankfurt/Germany
5. www.andritz.com/products-en/group/nonwoven-textile/
→ www.techtextil.com needlepunch/excelle-range/prodyn-forming-needlepunch-
nonwoven-and-textile
23-26 Texprocess 2024, Place: Frankfurt/Germany
6. https://bitbucket.org/hammerims/uniformity_index
→ www.texprocess.com