Technology for Wood and Natural Fiber-Based Materials

Research Project

Measurement of wood particle size – but how?

If fibres, wood chips and strands are to be used to produce board materials, it is necessary to determine their size and to define and characterise the frequency distribution of size parameters such as length and width. During manufacture, however, these characteristic values are not readily accessible and are, if at all, usually only determined through time-consuming sieve analysis. However, with the implementation of image processing methods, tailored to suit the respective materials, automatable laboratory and online measurements are now possible.

Optical measurement technology and process-oriented image processing offer numerous possibilities regarding the determination of particle size distribution. In the field of wood technology, however, the range of questions is extremely varied. Depending on the

  • material (fibres, wood chips, strands, singly or as fleece or on the board surface),
  • formulation of question (proportion of various particle size fractions, form factors),
  • answer time (laboratory measurements, measurement in component current, online monitoring) and
  • depth of information (screening curve, frequency distribution, statistical parameters such as modal values and quantile, deviation from a nominal distribution),

the correct strategy must be chosen for the preparation of material and the image acquisition and evaluation technology. At the WKI, particle size distribution is measured on single wood fibres and chips using the FibreShape system. In order to determine the chip size on the board surface or in the fleece, special image processing methods have been developed.

Examples and results

© Fraunhofer WKI
Beech chips (left) and spruce chips (right) from the same sieve fraction of 1.25mm – 3.15mm for the production of particle boards.
© Fraunhofer WKI
© Fraunhofer WKI

Differentiation between wood species-specific chip forms

When chips are produced from different species of wood, the resulting chip shape varies due to the anatomy of the wood. Although the differences are visible, they are not distinguishable through sieve analysis.

The top illustration shows beech chips (left) and spruce chips (right) from the same sieve fraction of 1.25mm – 3.15mm for the production of particle boards.

Measurement of the individual wood chips with the FibreShape device showed that the length distribution of the beech chips is wider than that of the spruce chips and that a substantial portion of the beech chips is longer than 1.5mm.

© Fraunhofer WKI
Beech fibres (TMP pulping with a refiner plate setting of 0.4mm) and on the right with a 1mm refiner plate setting.
© Fraunhofer WKI
© Fraunhofer WKI

Measurement of MDF fibres

MDF fibres resulting from refiner pulping are almost impossible to characterise through vibration screen fractionation, as they agglomerate in fibre balls. The top illustration shows beech fibres on the left (TMP pulping with a refiner plate setting of 0.4mm) and on the right with a 1mm refiner plate setting.

The fibre length distribution is determined using the FibreShape device. For this, individual fibres are placed upon a transparent film and are automatically measured by a flatbed scanner. Larger refiner plate settings during production in a refiner empirically result in a higher proportion of larger fibres and fibre bundles (shives). With a refiner plate setting of 1mm (lower illustration), the diagram of the measured length distribution clearly shows a fraction with lengths of around 10mm (104 µm). This fraction is missing when the refiner plates are set at 0.4mm (middle illustration).

© Fraunhofer WKI
Surface layer of a particle board (spruce chips, industry) Fig. left: grey image from the scanner (approx. 21 x 21mm²), right: detection results.
© Fraunhofer WKI
Measured length distribution for particles of > 0.5 mm²: individual measurement, image acquisition with FibreShape.
© Fraunhofer WKI
Measured length distribution for particles of > 0.5 mm²: measurement of 28 images of parts of the board surface.

Particle size measurement on the board surface

If the particle contingent cannot first be prepared and the outlay regarding the image acquisition needs to be minimised, complex image processing methods must be implemented. For scenes with non-isolated particles, the advanced layer analysis (see literature) is suitable. This treats the grey-scale images like a “grey-scale mountain range” and can restore particle contours. A particle size measurement directly on the board surface is, with certain restrictions, thereby possible.

The illustration top left shows a section of the surface of a laboratory worktop taken using a flatbed scanner; top right is the detection result. A sample of the un-pressed surface layer particles was additionally photographed and measured using the FibreShape device, whereby the fine material was sieved out in advance. The 1200dpi photos of the particle contours were then also divided into smaller parts and measured using standard methods. For both evaluations, particles with an area smaller than 0.5mm² and with form factors which indicated an irregular particle shape were not taken into account.

The measured distribution of particle length data was very similar for the two different methods. The proportion of longer particles is slightly higher for the individual measurement, but the modal values correspond entirely. When particle board surfaces are assessed, the “fine-particleness” also plays a role, i.e. the emergence of loose particles or other conspicuously large particles. This proportion is found in both distributions for particle lengths from around 5mm and can be assessed using e.g. the quantile. 

© Fraunhofer WKI
© Fraunhofer WKI
Length histogram as distribution density (blue) and distribution sum (green), obtained from approx. 60 images: using a flatbed scanner.
© Fraunhofer WKI
Length histogram as distribution density (blue) and distribution sum (green), obtained from approx. 60 images: taken from the mat.

 

Measurement of strands the mat

The illustration top left shows a scene of CSL strands similar to an OSB mat on the forming line. Top right is the result of the strand orientation through advanced layer analysis. When the strands are manually separated, photographed using the flatbed scanner and their contours are measured using standard methods, the determined length distribution shows a clearly-defined modal value in accordance with the length predetermined by the flaker. For the measurement in the mat, however, only the visible contours are recorded for all partly-obscured particles. This therefore results in a length distribution which still includes a modal value, but also includes an additional proportion of “smaller” particles, as these were only partially visible. When the degree of orientation of the strands is the same, this distribution is only dependent on the particle length itself and is definitely suitable for the continuous monitoring of the particle geometry. 

 

Funding:

In-house research