4.2. Analytical methods for characterising the fractionation of pulp

4.2.1. Storage of pulp samples

It is not possible in an extensive experimental study to analyse pulp samples immediately in a fresh state, but rather they have to be stored in a cold room or freezer for varying lengths of time. It is conceivable, of course, that the fibre properties of mechanical pulp may change in time, e.g. due to extraction and biodegradation, but cold storage at +3 °C for up to six weeks seems to have no appreciable effects on the freeness, strength or optical properties of TMP (Stoor et al. 2000). Similarly, pulp properties do not change upon storage in a freezer, except that freeness may increase slightly.

A cold room at +3 °C was used to store the present pulp samples, and the duration of storage before analysis was up to two weeks. The effect of storage on pulp properties was thus considered to be negligible.

4.2.2. Consistency

Consistency is the most common parameter analysed in pulp slurries, because information on the dry matter content of the suspension is needed for almost all other pulp analyses. It is also one of the few properties of pulp which can be measured on line, although the reliability and accuracy of such measurement is dependent on the method used and the process conditions. Consistency is an essential parameter when studying screening, because together with volumetric flow measurements it provides an instrument for determining the mass reject rate.

In the recent investigation into changes in groundwood pulp properties within the screen basket (Ämmälä et al. 1999b), a considerable correlation was observed between consistencies and both freeness values and strength properties. It was thus concluded that consistency could be used as a means of quality control and monitoring in pressure screening.

Manual consistency measurement can be considered one of the most accurate pulp analyses, although considerable errors can occur even here. The coefficient of variation in determining the consistency of mechanical pulps of low consistency (0–4%) has been measured in our laboratory to be less than 1%.

4.2.3. Freeness

The freeness test, which measures drainability, is one of the most traditional methods for characterising pulp. The most common tester for mechanical pulps, the Canadian Standard Freeness (CSF) tester, was introduced in 1925 and became the first standardised pulp test for general use in newsprint mills (Clark 1985). As measurements are quick and easy to perform, it is still in common use in paper mills to predict the quality and dewatering tendencies of pulps. It has often been criticised, however, because it does not describe explicitly the effective dewatering of pulp on the wire section of the paper machine (Clark 1985). Freeness is certainly not a universal measurement of pulp, but rather a process-specific one. Corson (1996) has suggested that in the case of a given mechanical pulp, freeness is probably the best indicator of its papermaking properties.

The Canadian standard method for the freeness test was originally designed for use with groundwood pulp, but it is also used for other mechanical pulps and even chemical pulps. It has been shown, however, that freeness is actually a measure of the specific surface of the fibres and does not respond very selectively to changes in their specific volume (El-Hosseiny & Yan 1980). The former correlates with the papermaking potential of mechanical pulp and the latter with that of chemical pulp, which is one reason why the method is well suited for mechanical pulps but not so appropriate for chemical pulps. The other reason is derived from observations that chemical pulps may be sensitive to the conductivity of the filtrate in freeness testing (Hiltunen 1999), whereas mechanical pulps are not (Ämmälä et al. 2000a).

The repeatability of freeness measurements is quite good. The study of freeness analysis by Ämmälä et al. (2000a) showed that a laboratory technician having good analytical skills and a good routine can attain a measurement value in a single freeness test which deviates by less than ±4% from that in the sample at a confidence level 95%. The relative accuracy seemed to be slightly dependent on the freeness level of the pulp, decreasing as the freeness level decreased. In the light of one earlier study (Ämmälä et al. 1999c), the most inaccurate step in the whole analysis procedure may be taken to be sampling of the process (Ämmälä et al. 2000a).

It has been noted in mechanical pulp testing that pulps from hot conditions that have been subject to grinding and refining do not exhibit their full papermaking potential and freeness if prepared in the cold. If hot fibres are cooled while in a curled and distorted state, the lignin sets and the fibres remain deformed (Jones 1966). The properties of such a pulp can be restored by means of hot disintegration. The condition in which some properties of a mechanical pulp are inhibited and require hot mechanical treatment for their development has been termed — latency — by Beath (1966).

Pulp which has once lost its latency may recover it during cold storage, and therefore its properties are usually different from those of fresh pulp. Consequently hot disintegration is often recommended in mechanical pulp testing. As disintegration can be considered an extra working stage in the analysis procedure and an extra drain on the resources of the laboratory, the need for disintegration was studied in screening experiments. The results showed (Ämmälä et al. 2000a) that even if hot disintegration alters the freeness value of a pulp considerably, the relative freeness change for pulp samples after the latency tower remains constant (see Fig. 8). After hot disintegration at 85 °C, the freeness for TMP pulps was about 86% of the value for non-disintegrated pulp, the correlation coefficient being 0.99. The disintegration consistency (0.4–3.0%), freeness level (30–600 ml) and origin of the pulp (TMP from two different mills) seemed not to have any systematic effect on the relative drop in freeness during disintegration. Temperature seemed to be a critical parameter, for if too low a temperature was used (73 °C), latency was abolished incompletely, giving a lower freeness drop. Cold disintegration (20°C) gave only minimal changes in freeness compared with non-disintegrated pulp.

It was concluded that the disintegration of pulp is not necessary for the analysis of freeness in samples from screening experiments, and the step was left out of the analytical procedure employed here.

4.2.4. Bauer-McNett classification

The Bauer-McNett classification is a much used method for characterising both chemical and mechanical pulps, although it is more suitable for the latter, because the fibre length distribution is usually broader than in chemical pulps and thus the fibres will be distributed more evenly in each fraction. It has been shown (Lindholm 1980, Koran 1994) that these fractions serve well to predict most of the properties of a given mechanical pulp. Hence it seems to be a suitable method for the analysis of pulp fractionation. As there are several ways of representing Bauer-McNett fractions, the following notations are selected for use here for the sake of clarity:

• R16 is the fraction that remains on the 16-mesh (1190 µm) wire.

• R30 is the fraction that remains on the 30-mesh (595 µm) wire but passes through the 16-mesh wire.

• R50 is the fraction that passes through the 30-mesh wire but remains on the 50-mesh (297 µm) wire.

• R200 is the fraction that remains on the 200-mesh (74 µm) wire, but passes through the 50-mesh wire.

• P200 is the fraction that passes through the 200-mesh wire.

The Bauer-McNett apparatus has been shown to classify pulp primarily according to fibre length, while other factors such as width and fibre flexibility have only minor effects (Ullman et al. 1968). The analogy between pressure screening and Bauer-McNett classification is obvious. The analysis as such has been shown to be repeatable (within the same laboratory) but its reproducibility (between laboratories) has been found to be poor (Levlin 1982). The repeatability, expressed as the coefficient of variation, has been reported to be of the order of 1% for each fraction (Ullman et al. 1968, Levlin 1982). The poor reproducibility is attributable to dimensional and constructional anomalies in the apparatus (Bos 1966), and the differences between the results of analyses performed with different apparatuses have been shown to be so great that attempts at developing Bauer-McNett analysis into a standard method have been abandoned (Levlin 1982).

Although individual analyses are not directly comparable between apparatuses, a correlation seems to exist between them. An example of the results obtained here with two apparatuses using the same TMP samples are presented in Fig. 9. As can be seen, they are quite divergent, but an almost linear relationship exists between the individual fractions determined with these two apparatuses, as shown in Fig. 10. Thus it can be assumed that different Bauer-McNett apparatuses produce comparable results in pressure screening if the normalised values (i.e. the output value divided by the input value) is considered, even though Bauer-McNett classification results for single streams are not directly comparable. This is an important observation, because it makes it possible to compare fractionation analyses performed with Bauer-McNett apparatuses producing very different classification results. The assumption may not be always valid, however, as the interrelation between apparatuses is not necessarily linear and the lines describing the results do not necessarily intersect at the origin.

4.2.5. Modelling of Bauer-McNett fractions

Bauer-McNett classification and the optical fibre analyser actually measure the same fibre property, length, and its distribution, and a linear relationship has been found between fibre lengths measured by the two methods (Bentley et al. 1994). From the very beginning researchers have tried to model Bauer-McNett fractions by means of fibre length distribution (Tiikkaja 1996), but the results have not been good enough and the predictive power of the models has been shown to be limited. In the case of mechanical pulps, one disadvantage of an optical analyser is poor resolution in the case of fines, the resolution of the optical analyser as a whole being about 50 µm, whereas most of the particles in the P200 fraction are smaller than this. Taking this into account, a model was developed in which correction factors were used. These were determined for each fraction, those for the fines (P200) and for the coarsest material (R16 for TMP or R30 for GW) being dependent on the freeness level while the others were constant over the whole freeness range. The principle of the modelling was introduced by Niinimäki et al. (1999).

The model was shown to be accurate if correction factors were determined for a given pulp in a given process, and in view of its versatility and good prediction ability, this model was chosen as the method for this thesis. Modelled and actual Bauer-McNett fractions were verified with independent test data from time to time during the project, with the results depicted in Fig. 11. All the proportions of the fractions by weight are plotted on the same figure. As can be seen, the accuracy of the model is very satisfactory.

4.2.6. Optical fibre length and coarseness

The first mill installation for the optical analysis of shive content dates back to the mid-1970’s (Sköld & Nilsson 1993), while the first commercial optical fibre analyser was introduced in the early 1980’s (Tiikkaja 1997), after which they quickly became common in pulp and paper laboratories. Optical analysers had great advantages over manual methods because of their ease and quickness of use.

The device used in this work was mostly the Kajaani FSA fibre length analyser, which allows only the fibre length distribution to be measured. Towards the end of the study, however, the FSA was replaced with the Kajaani FiberLab for the analysis of samples from mill screening experiments. The principle of fibre length measurement in both Kajaani devices is in accordance with the TAPPI T271 standard, and the repeatability of the measurement has been shown to be very good (Luukkonen et al. 1990). In addition to fibre length distribution, FiberLab also allows the coarseness, width and wall thickness of fibres to be analysed.

The coarseness of fibres, ω , can be measured easily with optical fibre analysers. It can be calculated by dividing the weight of the analysed sample by the cumulative fibre length:

Equation 21.

where

n is the number of fibres analysed,

l_a is the arithmetic mean length of fibres (m),

m_s is the weight of fibres in the sample (mg).

The length resolution of the optical fibre analyser, about 50 µm, is inadequate for the smallest particles in wood pulps, and this situation is emphasised further in the case of mechanical pulps, as these contain large amounts of fines, normally in the range 10–45 w-%. Thus coarseness measured directly by the analyser is not usually very useful as such, since the fines have a notable effect on the weight of the sample but not on the fibre length. Consequently some correction is needed. This was done here by multiplying the coarseness value by the weight proportion of fibrous material, referring to the new quantity as the corrected coarseness:

Equation 22.

where

ω ´ is the corrected coarseness (mg/m),

ω is the coarseness measured with the optical fibre analyser (mg/m),

w_fines is the weight proportion of fines (Bauer-McNett P200 fraction).

The proportion of the P200 fraction by weight was chosen for correction because the wire opening (74 µm) is close to the resolution of the analyser (50 µm) so that the majority of the particles in this fraction are not measurable optically. If one takes into account that the R200 fraction also contains some material that is too small to be measured, the weight of the P200 fraction seemed to be a reasonable approximation for the total amount of unanalysed particles in optical fibre length analysis.

4.2.7. Coarseness index by image analysis

FiberLab employs a CCD camera and two-dimensional image analysis, making it possible to determine distributions and average values for fibre length, width and cell wall thickness. About 10% of the total number of fibres in the sample are analysed by image analysis. Fibre width and cell wall thickness can be used to determine the coarseness index, CI, which is assumed to be proportional to the cross-sectional surface area of a fibre. The coarseness index is calculated from (Kauppinen 1998):

Equation 23.

where

W is the average width of the fibres analysed (µm),

CWT is the average cell wall thickness of the fibres analysed (µm).

Since cell wall thicknesses measured with FiberLab are generally much greater than those measured by light microscopy or CLSM (Heikkurinen et al. 1998), the FiberLab values can be considered to express relative wall thicknesses rather than actual values.

If the cell wall density is assumed to be constant for every measured fibre, a linear relation should exist between coarseness (in its corrected form, see Eq. 22) and the coarseness index. To explore the interrelation between these two methods, the results of analyses of pulp samples from two separate mill screenings were drawn in same figure (Fig. 12). As can be seen, the correlation is in general quite good. The data are from test series 1 and 3 (see section 4.5), which were analysed in different laboratories, with different settings of their FiberLab devices. The discrepancy between the coarseness indices is mainly attributable to the fact that fibres longer than 0.2 mm were subjected to image analysis in series 1, while only those over 0.4 mm were analysed in this way in series 3. Part of the discrepancy may also be due to differences in refining process conditions between the two mills.

The FiberLab coarseness index based on the cross sectional area of fibres determined by fibre width and cell wall thickness was found to be an expedient method for expressing the coarseness of TMP fibres. The analysis is quick and easy to perform, and it seems to give results that are almost directly proportional to the coarseness of the pulp after removal of the fines (Bauer-McNett P200 fraction), i.e. the corrected coarseness as defined here. The coarseness index may not be a universally explicit function of the actual coarseness of fibres, but they probably have a mutual pulp and process-specific correlation. Due to the use of fibre splitting as an analytical method, incompletely peeled fibre wall material, external fibrillation and changes in specific volume (internal fibrillation) may cause variations in the result. The middle lamella, for instance, does not peel away completely in refining, and the middle lamella coverage in TMP fibres has been measured to be about 15–25% (Fjerdingen et al. 1997). Fibre splitting is closely dependent on the pulping method, its incidence in the Bauer-McNett fraction greater than 48-mesh having been determined to be about 10% for TMP, but as much as 35% for PGW (pressure groundwood) and 46% for GW pulp (Fjerdingen et al. 1997). It can therefore be assumed that the coarseness index determined by image analysis is a practicable method for relative intact fibres such as those in TMP but may give greater variation and lower correlation with the actual coarseness of fibres in the case of damaged fibres such as those in GW. This was not assessed here, however. In general, the reliability of the analytical results can be ascertained by analysing a sufficient number of fibres, which means in practice that several parallel analyses must be performed on the same pulp sample.

4.2.8. Summary of laboratory analyses

The analytical methods evaluated and developed for this fractionation study are listed in Table 2.