Precision and Accuracy Studies with Kajaani Fiber Length Analyzers
The aim of this study was to test the measurement precision
and accuracy of the Kajaani FS-100 giving attention to possible machine
error in the measurements. Fiber length of pine pulps produced using polysulfide,
kraft, biokraft and soda methods were determined using both FS-100 and
FiberLab automated fiber length analyzers. The measured length values
were compared for both methods. The measurement precision and accuracy
was tested by replicated measurements using rayon stable fibers. Measurements
performed on pulp samples showed typical length distributions for both
analyzers. Results obtained from Kajaani FS-100 and FiberLab showed a
significant correlation. The shorter length measurement with FiberLab
was found to be mainly due to the instrument calibration. The measurement
repeatability tested for Kajaani FS-100 indicated that the measurements
Methods developed to determine fiber dimensions could be divided into
manual and automated methods. Based on measurement methods, fiber dimensions
can be determined either by direct measurements (e.g., microscopic counting,
image analysis, measurement from slurry) or by an indirect fractionation
method. Until recently, fiber length and coarseness were measured using
the method explained in TAPPI Method T234, which is lengthy, but may be
performed rapidly to determine fiber coarseness and length (Butler, 1948;
Reed and Clark, 1950; Anderson, 1953). Today modern image analyzers make
it easy to measure fiber length and to define fiber coarseness in a short
time because optical analyzers are faster and less tedious. Moreover,
recent improvements in computer-based image processing systems have led
to the development of fiber analyzers that provide rapid and operator-independent
The earliest method to measure the fiber length, width and coarseness
was the microscopic method (Wilson, 1954). This procedure suffered from
defects (fractionation of flat or collapsed fibers) that occur during
measurements on the fibers. In addition, every single fiber in the specimen
had to be measured to the nearest 0.1 mm and experiments had to be performed
meticulously, which took a very long time to complete (Clark, 1942). Instead
of the microscopic method, the projection method makes fiber length measurements
easier (Sugden, 1968). In this method, a fiber photo is projected on to
a surface and dimensions are measured manually. In addition, it is possible
to measure the fiber widths to the nearest micron with this technique,
but of course measurements take time and require careful attention. When
weight-weighed average fiber length is of interest, problems arise, since
most fiber length measurement methods do not discriminate fibers according
to their weight. The weight-weighed average fiber length of a pulp sample
can be determined with conventional methods using screen classification
of a sample onto different length fractions using the Bauer McNett fiber
classifier. This separates the fibers into length classes by using a relatively
dilute fiber suspension. Four or more wire-mesh screens in compartments,
each screen having successively smaller openings separates fiber based
mainly on length characteristics (Nadelman et al., 1955; Edgerton,
1956; Vecchi, 1969; Tasman, 1972). On the other hand, studies revealed
that several factors influence the results of Bauer McNett fractionation
such as the amount of water used in classification, screen size, sequence
of screens and etc. In addition, Bauer McNett classifies fibers not only
on the length base, but fiber coarseness and flexibility are also parameters
playing a role in classification. Conclusively, inferior reproducibility,
especially for chemical pulps, is the major problem for this measurement
method (Levlin, 1982).
Today, there are several automated optical fiber length analyzers on
the market. Kajaani introduced three types of laboratory fiber analyzers:
The Kajaani FS-100 and FS-200 (Young, 1993) and FiberLab and an online
analyzer (FSA) (Tiikkija, 1994). The first automated fiber analyzer, the
Kajaani FS-100 (Bichard and Scudamore, 1988) was introduced to the pulp
and paper industry in the 1980 and is an optical device accepted as method
for laboratory fiber length measurements (Tappi T271) to measure fiber
length and coarseness. This instrument is designed to make measurements
within minutes with one simple measurement procedure (Piirainen, 1985).
It consists of a capillary tube (0.2 mm) through which an aqueous suspension
of the fibers is passed. A light source is located on one side of the
capillary and a detector is on the opposite side. As a fiber passes through
the capillary tube, its polarized image is projected onto the detector,
which provides lengthwise dimension of the fiber, which can be calculated.
A low-pressure vacuum pump and chamber collect the analyzed fibers. The
measurement range is between 0-6.79 mm, divided into 24 classes, of which
the first 12 classes are resolved to 0.2 mm lengths and the last 12 have
a resolution of 0.4 mm (for the 0-0.7 mm range) (Jackson, 1988).
One of 3 measuring ranges is chosen previous to the sample is introduced.
As the diluted fiber suspension (0.0004% consistency) is introduced by
pouring it into the capillary funnel, fiber counting is initiated manually
on the keyboard. After all the fibers have passed through the capillary,
the counting is stopped manually at the keyboard and the results of the
fiber length measurement are printed out. The essential number of fibers
is counted and fiber length characteristics are presented, including the
cumulative fiber length distribution, arithmetic and length-weighed average
fiber length. In addition, the graphical presentation of results is obtained
directly from the FS-100 output to show fiber length distribution, both
in population terms and weight by results. As the Kajaani apparatus is
based on the measurement of the total projected length of a known amount
of pulp fibers with an optical image analyzer, coarseness is calculated
simply by dividing the pulp mass by the total measured length of the fibers.
The Kajaani FS-200 is designed to be a process control instrument and
is wholly automated; therefore, the capillary tube diameter changes from
0.2 -0.4 mm in FS-200 automatically. Providing enough vacuums ensures
the fibers in the capillary tube to be straight. A laser light source
is used and on the opposite side, a highly sensitive detector is set up
(Luukkonen et al., 1990).
Kajaani introduced the new laboratory fiber analyzer, the Kajaani FiberLab,
in 1997 (Tiikkaja, 1997). This new device measures fiber length by the
same method as used in the FS-100 and FS-200, but in addition it measures
fiber width and cell wall thickness. This new device is easy to use, fully
automated and modular. FiberLab uses both conventional optical measurements
and 2D-image analysis techniques developed especially for this device.
Measurements take place in a capillary tube, one fiber at a time, where
fiber length is measured according to TAPPI method T-271 with 50 μm
resolution and thus, fiber width and cell wall thickness is measured from
the same fiber (Tiikkaja, 1997). FiberLab measures thousands of fibers
in only 8 min and determines fiber width and cell wall thickness with
1 μm resolution. It is able to determine the ratio of three different
wood species in a blend from the fiber length, width and cell wall thickness
Several other optical fiber length analyzers are on the market today,
such as the Galai CIS-100 and SilviScan-2, the Imaging Fiber Analyzer
(Olson et al., 1995) and the Confocal Laser Scanning Microscopy
(Jang et al., 1992). Although the CIS-100 system is a particle
analyzer based on the direct visualization of the fiber and it also measures
fiber coarseness. In contrast to the FS-200, the CIS-100 is not restricted
just to the cellulosic fibers and thus is more generally applicable. The
SilviScan-2 is an instrument that is combined with a scanning x-ray micro-densitometry,
x-ray diffractometry and image analysis to measure wood fiber properties,
which in addition controls the fiber quality. Fiber diameter, wall thickness,
coarseness and wall perimeter, specific surface, wood density and microfibril
angle can all be easily measured with this system. The Imaging Fiber Analyzer
determines the distribution of fiber shape and length in samples of papermaking
fibers using a CCD camera. Confocal laser microscopy in combination with
image analysis allows determining the collapse and coarseness of fibers
in paper sheets from cross-sectional dimensions (Sundin et al.,
Kajaani FS-100 is in operation measuring the length of fibers. One measurement
error, a machine error for Kajaani FS-100 has not drawn much attention.
Therefore, the objective of this study is to evaluate Kajaani FS-100 for
measurement precision and accuracy and compare the measured values with
MATERIALS AND METHODS
Unbleached pulp samples from Scots pine (Çöpür, 2003),
which was produced using different alkaline pulping methods, polysulfide,
kraft, biokraft and soda, were used for fiber length and coarseness analysis.
Kajaani FS-100 fiber length analyzer was used to measure the numerical
average fiber length and the fiber coarseness values from the never-dried
screened pulps. Coarseness was determined by dividing the pulp mass (oven
dry) by the total measured length of the fibers. All measurements were
performed in five replication and the instrument was calibrated prior
to use by running a sample of synthetic (rayon) calibration fibers. Fiber
length determinations were achieved with the system optics adjusted to
the 0-7.0 mm measuring range. Metso Automation at Kajaani measured never-dried
pine samples by using the Kajaani FiberLab from screened pulps. Twenty
replicate measurements with rayon stable fibers were made to determine
the Kajaani FS-100 precision. The repeatability of the Kajaani FS-100
was also tested with kraft fibers collected from the compartment of Bauer
McNett classifier screen 14.
RESULTS AND DISCUSSION
In general, the Kajaani FS-100 instrument is considered to generate repeatable
results, as indicated by small variations observed between replicate measurements.
However, for the present study verifying the precision, accuracy and repeatability
of the FS-100 measurements was appropriate. A rayon staple fiber, provided
by the equipment manufacturer, was used for these experiments. Runs were
made using new aliquots from a single preparation to provide 20 replicate
measurements to test the precisians. The average length determined was
0.8836 mm with standard error of 0.0016 at a 95% confidence level. This
indicated that measurements with Kajaani FS-100 are precise (Fig.
||Repeatability control for Kajaani FS-100 with staple
Repeatability in an instrument can be measured if the same sample is
measured several times in a recycling manner. In this study the sample
was collected after the first run and rerun for 4 additional times. Table
1 shows the results obtained after each measurement and results indicated
that after first run, fibers were broken into parts. This is most likely
a result of the sudden pressure drop at the exit from the measurement
tube. This effect continued throughout the additional cycles. The arithmetic
average fiber length became smaller with an increasing amount of fines
and fiber counts as the fibers were reanalyzed. Therefore, this procedure
did not allow us to determine the repeatability on Kajaani FS-100 to determine
the machine error. The results from the precision study, which include
an aliquot sampling error, do set an upper limit for the repeatability
standard error of 0.0016 at 95% confidence interval. This overestimate
of the repeatability error would indicate a high enough repeatability
to have confidence in the instruments results, even though it could not
be directly measured.
Table 2 gives the fiber length averages, arithmetic,
L(n) length-weighed, L(l) and weight-weighed, L(w) for the five replicate
determinations for pine pulps. In general, each instrument was found to
generate repeatable results as indicated in small variations observed
between replicate measurements.
Fiber tests that give numerical values should be relatively independent
of the method or instrument used. It may not be possible to get a similar
value in one test using a different instrument, but the values should
still correlate. Automated optical fiber analyzers make it possible to
get reliable fiber length data in a relatively short period of time. However,
the differences between instruments have an important effect on the results
for average fiber length.
Basically, the differences between the Kajaani FS-100 and FiberLab fiber
length analyzers are the method and accuracy of measurement. A brochure
from Metso Automation Kajaani Ltd (2000) reveals that FiberLab measures
fiber cell wall thickness and fiber width with 1 μm and length with
50 μm resolutions. The resolution with the Kajaani FS-100 is lower
because the Kajaani FS-100 is able to measure at 24 classes, 12 of which
have resolution at 0.2 mm and the remaining 12 have a resolution of 0.4
||Repeatability control for Kajaani FS-100 with the same
pulp fibers in a recycling manner
||Fiber length analysis of pine fibers measured with FS-100
In addition, the Kajaani FS-100 does not include the first and last fiber
length classes in the calculation, which means that a significant number
of fines in the first class will not be included in the average calculation.
In addition, it is possible to measure around 20,000 fibers with FiberLab,
which should yield statistically more accurate fiber length data.
A comparison with Kajaani FS-100 revealed that the FiberLab measured
the same fibers on average 0.142 mm shorter (range 0.09-0.19). Finer sensitivity
and better resolution of FiberLab detected more short fibers and fines
and revealed 1.8% more fine contents. Therefore the regression results
showed that arithmetic average fiber lengths (R,% = 91.7-L(n)), fines
content (R, % = 72.8) and coarseness (R, % = 83.8) measured with Kajaani
FS-100 and Fiberlab methods were well correlated.
The results indicated that the measurements using Kajaani FS-100 and
FiberLab were well correlated. FiberLab, the new technology, has the advantage
of measuring thousands of fibers in a short time without plugging. In
general Kajaani generated repeatable results. Replicate measurements showed
small variations. The standard error observed was 0.0016 at a 95% confidence
level. This indicated that measurements with Kajaani FS-100 are precise.
Anderson, O., 1953. An investigation of the hillbom and bauer mcnett fiber classifiers. Svensk Papperstid, 56: 704-709.
Bentley, R.G., P. Scudamore and J.S. Jack, 1994. A comparison between fiber length measurement. Methods Pulp Paper Mag. Canada, 95: 41-45.
Bichard, W. and P. Scudamore, 1998. An evaluation of the comparative performances of the Kajaani FS-100 and FS-200 fiber length analyzers. Tappi J., 71: 149-155.
Butler, W.T., 1948. The use of the bauer mcnett fiber classifier. Pulp Paper Mag. Canada, 49: 133-136.
Clark, J.D.A., 1942. The measurement and influence of fiber length. Tappi Techonol. Assn. Papers, 25: 556-562.
Copur, Y., D. Kiemle, A. Stipanovic, J. Koskinen and H. Makkonen, 2003. HNMR spectroscopic determination of carbohydrates and yield in pine and maple pulps. Paperi Puu, 85: 158-162.
Direct Link |
Edgerton, D.S., 1956. An evaluation of a grid technique for the determination of fiber length distribution in chemical pulps. Tappi, 39: 171A-177A.
Jackson, F., 1968. Fiber length measurement and its application to paper machine operation. Appita, 41: 212-216.
Levlin, J.E., 1982. On the suitability of the mcnett classifier for the fiber length classification of pulps. Paperi Puu, 64: 213-236.
Luukkonen, M., H. Suutari and L. Paavilainen, 1990. Utilization of pulp-drainage and fiber-length measurements. Appita, 43: 213-216.
Nadelman, A.H., R. Schmut, D.J. Kradske and R.I. Larson, 1955. The interrelationship of fiber length distribution, fiber length index and physical characteristics of handsheets. Tappi, 38: 304-309.
Olson, J.A., A.G. Robertson, T.D. Finnigan and R.H. Turner, 1950. An analyzer for fiber shape and length. J. Pulp Paper Sci., 21: J367-J373.
Piirainen, R., 1985. Optical method provides quick and accurate analysis of fiber length. Pulp Paper, 59: 69-71.
Reed, A.E. and J.A. Clark, 1950. An instrument for rapid fractionation of pulps. Tappi, 33: 292-298.
Sugden, E.A.N., 1968. A semi-automatic method for the determination of average fiber length by projection. Pulp Paper Mag. Canada, 69: 71-76.
Sundin, J., L. Sundsrom and N. Hartler, 1966. Determination of collapse and coarseness of fibers in paper sheets from cross-section dimensions using confocal microscopy and image analysis. Nordic Pulp Paper Res. J., 11: 56-57.
Tasman, J.E., 1972. The fiber length of bauer mcnett screen fractions. Tappi, 55: 136-138.
Tiikkaja, E., 1994. Experiences with a new on-line fiber-length analyzer. Pulp Paper Mag. Canada, 95: 207-209.
Tiikkaja, E., 1997. Kajanai fiber lab. Valmet Paper News, pp: 33-34.
Vecchi, E., 1969. Quality control of poplar groundwood: Factors related to the structural composition of the pulp. Tappi, 52: 2390-2399.
Wilson, J.W., 1954. Fiber technology; fiber length menstruation; a comprehensive history and new method. Pulp Paper Mag. Canada, 55: 84-89.
Young, S.C., 1993. A calibration and maintenance schedule for the kajaani FS-200 fiber analyzer. Appita, 46: 435-438.