PEER-REVIEW ARTICLE PEER-REVIEWED ARTICLE bioresources.com Ushakov et al. (2023). “Pulp properties & paper,” BioResources 18(4), 8090-8103. 8090 Factor and Cluster Analyses of the Structure of Correlations between High Consistency Pulp Properties during Refining and Paper Strength Characteristics Alexander Ushakov,* Yuri Alashkevich, Robert Pen, and Viktor Kozhukhov This article analyses high-consistency pulp refining using a disk refiner. During the experiment, the size of the gap between the rotor and stator cutters (0.5 to 1.5 mm), rotor speed (2,000 to 2,500 rpm), pulp consistency (10 to 20%), and freeness value (15 to 60 °SR) of the pulp were varied. The refining results were characterised by changes in 10 output parameters: morphological properties of cellulose fibres (average length, width, fibrillation index, water retention value, average kink angle, and coarseness) and the physical and mechanical characteristics of handsheets (breaking length, bursting strength, tearing resistance, and folding endurance). A total of 56 observations were made on the samples. Factor and cluster analysis methods were used to study the structure of correlations between the output parameters. More than 96% of the total dispersion of all output parameters was due to a change in two latent (hidden) factors: the first one was responsible for 79.6% of the dispersion and is presumably identified as the degree of external fibre fibrillation and the second one (16.6% of the dispersion) as fibre flexibility (including coarseness and average kink angle). DOI: 10.15376/biores.18.4.8090-8103 Keywords: Pulp refining; Correlation factor analysis; Cluster analysis; Morphology of cellulose fibres; Physical and mechanical characteristics of handsheets Contact information: Reshetnev Siberian State University of Science and Technology 31, Krasnoyarskiy Rabochiy Prospekt, Krasnoyarsk, 660037 Russian Federation; * Corresponding author: al.ushakov2194@mail.ru INTRODUCTION Pulp refining in disk refiners is an important operation in pulp and paper production on which the morphological properties of fibres and the strength characteristics of paper largely depend (Alashkevich et al. 2006, 2010; Chen et al. 2016). There are usually correlations between these characteristics. This indicates the existence of a smaller number of more general, “deep” properties of cellulose fibres that change during refining, which, in turn, results in a change (variance) in the measured characteristics of pulp and paper (Lawley and Maxwell 1962; Pen 1972; Everitt et al. 2011; Almonti et al. 2019). Such properties are called hidden (latent) factors. They can be identified and analysed by multivariate mathematical statistics, in particular, by the factor and cluster analysis methods (Pen 1972; Kim and Mueller 1986; Strand 1987; Brown et al. 2004; Pulkkinen et al. 2010; Novoselskaya et al. 2019). In this article, both these methods are used in the analysis of high-consistency pulp refining (from 5% and higher). This research is relevant because fibres during refining can experience various kinds of deformations and structural changes that, when compared to PEER-REVIEWED ARTICLE bioresources.com Ushakov et al. (2023). “Pulp properties & paper,” BioResources 18(4), 8090-8103. 8091 the refining of low-consistency fibrous suspensions, have some differences (Wathén 2006). These structural changes are characterised primarily by such important indicators as: external fibre fibrillation, which increases the outer surface characteristics of fibres and the number of interfibre bonds, as well as internal fibrillation, which is accompanied by an increase in interfibre bonding forces and flexibility and a decrease in fibre coarseness, without weakening the strength of the fibre itself (Matveev 1974; Bhardwaj et al. 2004; Hou et al. 2011; Chen et al. 2017). The listed properties of fibres are largely influenced by the consistency of the refined pulp. Increasing pulp consistency during refining ensures greater external and internal fibre fibrillation, thereby increasing the strength characteristics of the finished paper products (Kang and Paulapuro 2006; Lebedev et al. 2018; Przybysz et al. 2020; Penkin et al. 2022). A number of studies have shown that a high-consistency pulp refining process is appropriate for obtaining highly-extensible paper, which is especially important in the manufacture of sack papers (Henderson et al. 1965; Gurnagul et al. 2005). It is also worth noting that high-consistency pulp refining ensures the preservation of the original fibre length due to high interfibre friction and a decrease in the cutting action from the cutting edges due to relatively large gaps between the grinding surfaces of the rotor disks and the refiner stator (Fernando et al. 2012; Kerekes 2015; Ushakov et al. 2020). This high interfibre friction during high-consistency refining, on the one hand, ensures better fibre processing, giving fibres a high water retention value and an external specific surface area, i.e., indicators characterising changes in internal fibre fibrillation (Sundström et al. 1993; Fernando et al. 2007, 2011). On the other hand, an excessive increase in pulp consistency reduces fibre fibrillation, causing fibres to twist (Hartler 1995). This deformation is called “fibre latency” (Klark 1983; Gard 2002). When refining high-consistency pulp, fibre latency occurs due to mechanical effects in the refining zone of the disk refiner. This results in a large number of highly deformed (twisted, kinked, or crumpled) fibres in the pulp (Page et al. 1985; Pen and Karetnikova 2008). The purpose of this study is to identify the number and physical nature of hidden factors that cause the dispersion of morphological and paper-forming properties during high-consistency pulp refining. EXPERIMENTAL Materials Bleached LS -1 hardwood sulphate pulp (a semi-finished product from Ilim Group, Bratsk (Russia)) was selected for experimental studies. The degree of delignification (Kappa number) was 2.0 to 4.0. Before refining, the pulp was defibrated with water according to ISO 5263-3 (2004) standard. The refining was carried out in a laboratory disk refiner as presented in Fig. 1. Pulp of the required consistency was placed into hopper 1 of the disk refiner. After that, screw feeder 2 was used to transfer the pulp from hopper 1 to working area 3 to refine it between the cutters of rotor 4 and stator 5. Next, the pulp was passed through outlet 6. The gap between the rotor and stator cutters was changed using mechanical adjusting device 9 by moving the stator along its axis. The rotational speed of the disk refiner rotor drive and the screw feeder was regulated using SMV frequency converters (AC Technology Corporation, Lenze AC Tech, Uxbridge, MA, USA). PEER-REVIEWED ARTICLE bioresources.com Ushakov et al. (2023). “Pulp properties & paper,” BioResources 18(4), 8090-8103. 8092 Fig. 1. Disk refiner (1 – pulp hopper; 2 – screw feeder; 3 – refining zone; 4 – rotor disk; 5 – stator disk; 6 – outlet; 7 – screw feeder worm gear; 8 – electric motor of disk refiner rotor drive; 9 – mechanical adjusting device) To assess the variability of the pulp refining quality, an increase in the freeness value was determined according to the Shopper-Riegler method (°SR) using the SR-2 device (Manufacturer Metrotex, Moscow, Russia) as per ISO 5267-1 (2000). For analyses, pulp samples were taken with a freeness value of 15, 30, 45, and 60 °SR. According to ISO 16065-2 (2019), the morphological properties of fibres were determined at least three times for each sample using the MorFi Neo fibre analyser (Manufacturer “TECHPAP”, Gieres, France). The studied morphological properties of fibres included such indicators as listed below. Average fibre length L avg. (mm) was determined according to the Eq. 1, N L L i avg  =. (1) where Li is the length of developed fibres. As shown in the calculation diagram of Fig. 2, Li is determined by the fibre analyser as the sum of segments of rectilinear sections of fibres along their axis; N is the number of recognised fibres. Fig. 2. Fibre length recognition by fibre analyser (Li = FA + FB + FC + …. + FH + … + FP ) PEER-REVIEWED ARTICLE bioresources.com Ushakov et al. (2023). “Pulp properties & paper,” BioResources 18(4), 8090-8103. 8093 The fibrillation index Fib (%) was calculated as the ratio of the sum of the lengths of all fibrils to that of the lengths of all recognised fibres and can be expressed through the Eq. 2, %100 1 1 =   = = N i i N i i L F Fib (2) where Fi is the -sum of all fibrils per fibre, and Li is the length of developed fibres. Fibre coarseness k (mg/m) is calculated as the ratio of the mass of all fibres (recognised by the fibre analyser) to their total length. Average fibre width Z (µm), as well as average fibre length, is summarily calculated using the device for each fibre segment. Average kink angle A (°), is determined as the points of abrupt change in the direction of the fibres where they can break. In addition to the listed morphological properties of the fibres, the water retention value (WRV) was evaluated as an indicator characterising the fibre swelling degree according to ISO 23714 (2014). The WRV of the pulp indicates the moisture remaining in it after centrifugation under certain conditions. Pulp centrifugation was completed using the MPW – 310 device (MPW Instruments, Warsaw, Poland). The moisture content of the pulp after centrifugation was determined by the difference in the mass of the sample before and after drying (%), and can be expressed using the Eq. 3, %100 − = wet drywet В WW WRV (3) where Wwet is the mass of wet fibres after centrifugation (g), and Wdry is the -mass of dry fibres (g). To determine the physical and mechanical characteristics, a sheet machine (Werkstoffprufmaschinen, Leipzig, Germany) was used to form handsheets. Before testing the physical and mechanical characteristics, the handsheets were conditioned under standard conditions. The physical and mechanical characteristics of the handsheets were evaluated according to the following indicators: − Breaking length in accordance with ISO 1924-2 (2008), using a RMB 30M tensile testing machine (Experimental Production Workshop, Moscow, Russia); − Bursting strength in accordance with ISO 2758 (2014), using an EC35 apparatus (TMI 13-6, Rotterdam, Holland); − Tearing resistance in accordance with ISO 1974 (2012), using a RB-1 device (Experimental Production Workshop, Moscow, Russia); − Folding endurance in accordance with ISO 5626 (1993), using the DRK111B Folding Tester (Shandong Drick Instruments Co., Ltd., Jinan, China) Methods Pulp refining and factor and cluster analyses During the experiment, the size of the gap between the rotor and stator cutters (with a range of variation 0.5 to 1.5 mm), rotor speed (2,000 to 2,500 rpm), pulp consistency (10 to 20%), and freeness value (15 to 60 °SR) of the pulp were varied. Table 1 presents the input technological factors of the refining process and the output parameters of the morphological properties of the pulp and the physical and mechanical characteristics of the PEER-REVIEWED ARTICLE bioresources.com Ushakov et al. (2023). “Pulp properties & paper,” BioResources 18(4), 8090-8103. 8094 handsheets. A total of 56 observations (refining modes) were made on the samples. The observations and their statistical characteristics are given in the Appendix (Tables S1 and S2). The results (statistical characteristics, correlation, factor, and cluster analyses) were mathematically processed using the Statgraphics Centurion XVI software product (free version). Table 1. Factors of the Refining Process and Output Parameters Factors of the Refining Process and Output Parameters Designation In formulas In tables and figures Factors of the refining process Rotor speed (rpm) n X1 Gap size (mm) s X2 Pulp consistence (%) C X3 Freeness value (°SR) Sh X4 Output parameters Average fibre length (mm) Lavg. Y1 Fibrillation index (%) Fib Y2 Fibre coarseness (mg/m) k Y3 Average fibre width (µm) Z Y4 Water retention value (%) W Y5 Average kink angle (°) A Y6 Folding endurance U Y7 Breaking length (m) L Y8 Bursting strength (kPa) Ра Y9 Tearing resistance (mN) Е Y10 RESULTS AND DISCUSSION The preliminary statistical analysis of the observations revealed correlations between most of the output parameters (Table 2). This is a consequence of a relatively small number of common properties – "hidden" (latent) factors (common factor, latent factor) f j that exert a greater or lesser influence on the output parameters and determine the structure of the correlation matrix. Table 2. Correlation Y1 Y2 Y3 Y4 Y5 Y6 Y7 Y8 Y9 Y10 Y1 Y2 Y3 Y4 Y5 Y6 Y7 Y8 Y9 Y10 1 -0.681 -0.217 -0.743 -0.749 0.334 -0.436 -0.703 -0.697 -0.698 – 0.681 1 0.059 0.813 0.705 -0.574 0.834 0.851 0.909 0.868 – 0.217 0.059 1 0.307 0.595 0.383 0.011 0.431 0.375 0.327 –0.743 0.813 0.307 1 0.844 -0.381 0.659 0.874 0.877 0.821 –0.749 0.705 0.595 0.844 1 -0.176 0.473 0.913 0.888 0.837 0.334 -0.574 0.383 -0.308 -0.176 1 -0.431 -0.317 -0.329 -0.408 –0.436 0.834 0.011 0.659 0.473 -0.431 1 0.657 0.731 0.641 –0.703 0.851 0.431 0.914 0.913 -0.317 0.657 1 0.972 0.927 –0.698 0.909 0.375 0.877 0.888 -0.329 0.731 0.972 1 0.943 –0.698 0.868 0.327 0.820 0.873 -0.408 0.641 0.927 0.943 1 PEER-REVIEWED ARTICLE bioresources.com Ushakov et al. (2023). “Pulp properties & paper,” BioResources 18(4), 8090-8103. 8095 This study used factor analysis and cluster analysis to identify the number and nature of hidden factors [1 through 5]. Factor analysis was used to derive a regression equation based on a matrix of correlations between output parameters. Linear regression is written as Eq. 4, ititiii flflflY ++++= 2211 , i = 1, …, m; m ˃ t, (4) where f j is the hidden factor “common (latent) factor”; t is the number of latent factors; m is the number of output parameters (in the studied case, m = 10); l it is the loading of the jth latent factor on the ith output parameter; i represents the residuals representing the sources of deviations affecting only Y i. Equation 4 expresses the basic hypothesis of factor analysis: the set of correlated variables Yi (i = 1, 2, ..., m) can be represented as a linear function of a smaller number of latent factors f j (j = 1, 2, ..., t) and a set of independent residuals i. Factor analysis of the results given in Table S1 was performed by the Minres method. The Varimax Rotation criterion was used for orthogonal transformation of factor loadings. Statistical significance with a confidence of at least 95% was established for two latent factors responsible for 96.19% of the total dispersion of all 10 output parameters, including 79.56% of the dispersion for the first factor and 16.62% for the second one (Table 3). Table 3. Factor Loading Matrix after the Varimax Rotation Variable Factor Loading Estimated Commonality Factor 1 Factor 2 Y1 -0.744 0.024 0.555 Y2 0.918 -0.380 0.987 Y3 0.369 0.801 0.777 Y4 0.908 -0.032 0.825 Y5 0.922 0.328 0.958 Y6 -0.383 0.619 0.529 Y7 0.693 -0.332 0.591 Y8 0.972 0.086 0.953 Y9 0.984 0.012 0.969 Y10 0.942 -0.027 0.888 Contribution from total variance (%) 79.56 16.62 96.19 Figure 3 shows a two-dimensional factor space with the output parameters under observation. The coordinates of the points are factor loadings (see Table 3). In addition to factor analysis, cluster analysis of the output parameters was performed using the Ward’s, Distance Metric Squared Euclidean method (Fig. 4). The classification procedures used are based on different data grouping methods. The object of factor analysis is the matrix of paired linear correlations between the output parameters. In cluster analysis, the basis for grouping is the geometric distances between the normalised values of the output parameters. Nevertheless, the results of the groupings turned out to be identical. This is confirmed by a visual comparison of Figs. 3 and 4. PEER-REVIEWED ARTICLE bioresources.com Ushakov et al. (2023). “Pulp properties & paper,” BioResources 18(4), 8090-8103. 8096 Fig. 3. Plot of the factor loading Fig. 4. Results of clustering by fibre properties The main part of the output parameters (Y2, Y4-, Y5, Y7, Y10) was grouped into a relatively dense cluster on the positive part of the coordinate axis of the first latent factor (Fig. 3). The nature of this factor can be identified with the external fibre fibrillation degree: the positive correlation of this property with the water retention value of the pulp, the fibrillation index, and the strength characteristics of the handsheets did not contradict the generally accepted prior information. The negative correlation between the average fibre length Y1 and other characteristics in the studied cluster was unexpected. A possible reason is because of a decrease in the length of the fibres during refining with a simultaneous increase in the external fibrillation degree. Figure 5 shows the relationship between these properties for one of the refining modes. Similar dependences are revealed in other refining modes within the pulp consistency range of 10 to 20%. The second latent factor affects the coarseness of the fibres Y 3 and their average kink angle Y6. Presumably, the nature of this factor is associated with internal fibrillation. PEER-REVIEWED ARTICLE bioresources.com Ushakov et al. (2023). “Pulp properties & paper,” BioResources 18(4), 8090-8103. 8097 High-consistency pulp refining is accompanied by an increase in internal fibrillation, resulting in increased flexibility and plasticity. The increased flexibility is accompanied by an increase in the number of kinked fibres and a decrease in their coarseness, as shown in Fig. 6. Fig. 5. The relationship between average fibre length on the fibrillation index (refining mode: rotor speed 2,000 rpm; gap between the rotor and stator cutters 1.5 mm; pulp consistency 10, 15, and 20%) Fig. 6. The relationship between average kinked angle of fibres on their coarseness (refining mode: rotor speed 2,000 rpm; gap between the rotor and stator cutters 1.5 mm; pulp consistency 10, 15, and 20%) 0.76 0.78 0.8 0.82 0.84 0.86 0.88 0.9 0.92 0.5 1 1.5 2 2.5 A v e ra g e F ib e r L e n g th (m m ) Fibrillation Index (%) 10% 15% 20% 15 °SR 30 °SR 45 °SR 60 °SR 127 128 129 130 131 132 133 134 135 136 0.1 0.12 0.14 0.16 0.18 0.2 0.22 0.24 A v e ra g e N u m b e r o f K in k s p e r F ib e r (° ) Fibre Coarseness (mg/m) 10% 15% 20% 30 °SR 15 °SR 45 °SR 60 °SR PEER-REVIEWED ARTICLE bioresources.com Ushakov et al. (2023). “Pulp properties & paper,” BioResources 18(4), 8090-8103. 8098 CONCLUSIONS 1. The study conducted between the morphological properties of the fibers and the strength characteristics of the paper by the methods of factor and cluster analysis showed that the share of the total dispersion of the studied parameters of the refining process is determined by two hidden factors. One of the factors is related to the degree of fibrillation of the fibers, and the other to their flexibility. The total share of the total variance of all observed indicators, due to the influence of two hidden factors, is 96.2% of their total variance. The first of the identified factors determines 79.6% of the variance of the variance in the observed indicators, the second 16.6%. 2. The effectiveness of "information extraction" by multivariate mathematical statistics (correlation, factor, and cluster analysis methods) was demonstrated using the example of studying experimental data on the pulp refining process, which is one of the most important production processes in paper technology. It was established that the external fibrillation of cellulose fibres makes a major contribution to the variability of the physical and mechanical properties of paper sheets during high-consistency pulp refining. ACKNOWLEDGMENTS This work was carried out under the State Assignment issued by the Ministry of Education and Science of Russia for the project: “Technology and Equipment for the Plant Biomass Chemical Processing” by the Plant Material Deep Conversion Laboratory (Subject No. FEFE-2020-0016). This work was performed using equipment of the Centre for Collective Use of the Krasnoyarsk Research Centre of the Siberian Branch of the Russian Academy of Sciences. We express gratitude to the staff of this centre for their assistance in our research. REFERENCES CITED Alashkevich, Yu. D., Kovalev, V. I., and Nabieva, A. A. (2010). Influence of the Headset Pattern in the Process of Grinding Fibrous Semi-finished Products: Monograph in Two Volumes. Volume 1, Siberian State Technological University, Krasnoyarsk, Russia. Alashkevich, Yu. D., Reshetova, N. S., and Gudovskiy, V. P. (2006). Theory and Design of Machines and Equipment of the Industry: Tutorial in Two Parts. Part 2, Siberian State Technological University, Krasnoyarsk, Russia. Almonti, D., Baiocco, G., Tagliaferri, V., and Ucciardello, N. (2019). “Artificial neural network in fibres length prediction for high precision control of cellulose refining,” Materials 12(22), article 3730. 1-13. DOI: 10.3390/ma12223730 Bhardwaj, N. K., Duong, T. D., and Nguyen, K. L. (2004). “Pulp charge determination by different methods: Effect of beating/refining,” Colloids and Surfaces A: Physicochemical and Engineering Aspects 236(1), 39-44. DOI: 10.1016/j.colsurfa.2004.01.024 PEER-REVIEWED ARTICLE bioresources.com Ushakov et al. (2023). “Pulp properties & paper,” BioResources 18(4), 8090-8103. 8099 Brown, T., Miles, K., McDonald, D., and Wood, J. (2004). “Multivariate analysis of seasonal pulp quality variations in a TMP mill,” Pulp & Paper Canada 105(10), 230- 234. Chen, T., Xie, Y., Wei, Q., Wang, X., Hagman, O., Karlsson, O., and Liu, J. (2016). “Effect of refining on physical properties and paper strength of Pinus massoniana and China fir cellulose fibers,” BioResources 11(3), 7839-7848. DOI: 10.15376/biores.11.3.7839-7848 Chen, T., Wu, Z., Wei, W., Xie, Y., Wei, Q., Wang, X., Hagman, O., and Karlsson, O. (2017). “Optimizing refining conditions of Pinus massoniana cellulose fibers for improving the mechanical properties of ultra-low density plant fiber composite (ULD_PFC),” BioResources 12(1), 8-18. DOI: 10.15376/biores.12.1.8-18 Everitt, B. S., Landau, S., Leese, M., and Stahl, D. (2011). Cluster Analysis 5th Edition, Wiley, King’s College, London, UK. Fernando, D. (2007). Ultrastructural Characterization (Morphological and Topochemical) of Wood Pulp Fibres; Effects of Mechanical and Kraft Processes, Doctoral Thesis No. 2007:39, Swedish University of Agricultural Sciences (SLU), Department of Forest Products/Wood Science, Uppsala, Sweden. Fernando, D., Muhic, D., Engstrand, P., and Daniel, G. (2011). “Fundamental understanding of pulp property development during different TMP refining conditions using the application of new Simons’ stain method and fibre surface ultrastructure,” Holzforschung 65(6), 777-786. DOI: 10.1515/HF.2011.076 Fernando, D., Muhic, D., Engstrand, P., and Daniel, G. (2012). “Surface and internal micro/ultrastructure of TMP fibres produced during high-intensity refining elucidate the development of pulp and paper properties,” Holzforschung 66(4), 467-475. DOI: 10.1515/hf.2011.175 Gard, J. (2002). The Influence of Fibre Curl on Shrikage and Strength Properties of Paper, Master’s Thesis, Lulea University of Technology, Lulea, Sweden. Gurnagul, N., Merng, Ju. S., and Shallhorn, P. (2005). “Optimizing high-consistency refining conditions for good sack paper quality,” in: Appita Annual Conference, Melbourne, Australia, pp. 379-386. Hartler, N. (1995). “Aspects on curled and microcompressed fibers,” Nordic Pulp & Paper Research Journal 10(1995), 4-7. DOI: 10.3183/npprj-1995-10-01-p004-007 Henderson, J. T., Barton, J. S., and Erfurt, H. (1965). “Effects of high consistency refining on the properties of the consolidated web,” in: Consolidation of the Paper Web, F. Bolam (Ed.), Transactions of the Cambridge Symposium, Vol. 2, Pira, UK, pp. 875-895. DOI: 10.15376/frc.1965.2.875 Hou, Q., Yang, B., Liu, W., Liu, H., Hong, Y., and Zhang, R. (2011). “Co-refining of wheat straw pulp and hardwood kraft pulp,” Carbohydrate Polymers 86(1), 255-259. DOI: 10.1016/j.carbpol.2011.04.040 ISO 1924-2 (2008). “Paper and board — Determination of tensile properties — Part 2: Constant rate of elongation method (20 mm/min),” International Organization for Standardization, Geneva, Switzerland. ISO 1974 (2012). “Paper — Determination of tearing resistance — Elmendorf method),” International Organization for Standardization, Geneva, Switzerland. ISO 16065-2 (2019). “Pulps — Determination of fibre length by automated optical analysis — Part 2: Unpolarized light method,” International Organization for Standardization, Geneva, Switzerland. PEER-REVIEWED ARTICLE bioresources.com Ushakov et al. (2023). “Pulp properties & paper,” BioResources 18(4), 8090-8103. 8100 ISO 23714 (2014). “Pulps — Determination of water retention value (WRW),” International Organization for Standardization, Geneva, Switzerland. ISO 2758 (2014). “Paper — Determination of bursting strength,” International Organization for Standardization, Geneva, Switzerland. ISO 5263-3 (2004). “Pulps - Laboratory wet disintegration,” International Organization for Standardization, Geneva, Switzerland. ISO 5267-1 (2000). “Pulps - Determination of drainability — Part 1: Schopper-Riegler method,” International Organization for Standardization, Geneva, Switzerland. ISO 5626 (1993). “Paper — Determination of folding endurance,” International Organization for Standardization, Geneva, Switzerland. Kang, T., and Paulapuro, H. (2006). “Effect of external fibrillation on paper strength,” Pulp & Paper Canada 107(7/8), 51-54. Kerekes, R. J. (2015). “Perspectives on high and low consistency refining in mechanical pulp,” BioResources 10(4), 8795-8811. DOI: 10.15376/biores.10.4.8795-8811 Kim, J. O., and Mueller, C. W. (1986). Factor Analysis: Statistical Methods and Practical Issues, Sage University Paper Series on Quantitative Applications in the Social Sciences, Series № 07-014, pp. 1-88. Klark, J. (1983). Pulp Technology (Science of Pulp and Paper, Pulp Preparation, Paper Processing, Test Methods), Lesnaya Promyshlennost, Moscow, Russia. Matveev, B. P. (1974). Study of the Main Factors of the Process of Refining Sulphate Unbleached Pulp in a Disc Mill at High Consistency, Ph.D. Dissertation, Leningrad Technological Institute of Pulp and Paper Industry, Leningrad, USSR. Lawley, D. N., and Maxwell, A. E. (1962). “Factor analysis as a statistical method,” Journal of the Royal Statistical Society. Series D (The Statistician) 12(3), 209-229. DOI: 10.2307/2986915 Lebedev, I., Kazakov, Y., Chukhchin, D., and Romanenko, K. (2018). “The role of fibers surface in development of papermaking properties of technical pulp in the refining process,” Khimiya Rastitel'nogo Syr'ya 2018(2), 207-216. DOI: 10.14258/jcprm.2018022248 Novoselskaya, O. A., Kolesnikov, V. L., Penkin, A. A., Solov'eva, T. V., Kashin, A. N., Nagornova, I. V., and Varepo, L. G. (2019). “Management the paper printing properties by methods of factor analysis,” Journal of Physics: Conference Series 1260, article ID 032031. DOI: 10.1088/1742-6596/1260/3/032031 Page, D. H., Seth, R. S., Jordan, B. D., and Barbe, M. C. (1985). “Curl, crimps, kinks and microcompressions in pulp fibres – their origin, measurement and significance,” in: Paper Making Raw Materials, Transactions of the 8th Fundamental Research Symposium, Oxford, England, pp. 183-227. DOI: 10.15376/frc.1985.1.183. Pen, R. Z. (1972). “Factor analysis of the results of the active experiment,” Zavodskaja Laboratorija (Journal Plant Laboratory) 1, 70-73. Pen, R. Z., and Karetnikova, N. V. (2008). Wood and Waste Pulp Technology, Manual, Siberian State Technological University, Krasnoyarsk, Russia. Penkin, A., and Kazakov, Y. (2022). “Changes in the morphological characteristics of secondary fiber obtained from wet-strength paper during gentle refining of recycled pulp. Part 1. Fiber characteristics,” Russian Forestry Journal 2022(5), 157-172. DOI: 10.37482/0536-1036-2022-5-157-172 Przybysz, P., Dubowik, M., Malachowska, E., Kucner, M., Gajadhur, M., and Przybysz, K. (2020). “The effect of the refining intensity on the progress of internal fibrillation PEER-REVIEWED ARTICLE bioresources.com Ushakov et al. (2023). “Pulp properties & paper,” BioResources 18(4), 8090-8103. 8101 and shortening of cellulose fibers,” BioResources 15(1), 1482-1499. DOI: 10.15376/biores.15.1.1482-1499 Pulkkinen, I., Fiskari, J., and Alopaeus, V. (2010). “New model for predicting tensile strength and density of eucalyptus handsheets based on an activation parameter calculated from fiber distribution characteristics,” Holzforschung 64(2), 201-209. DOI: 10.1515/HF.2010.030 Strand, B. C. (1987). “Factor analysis as applied to the characterization of high-yield pulps,” in: TAPPI Pulping Conference, Washington, D.C., USA, pp. 61-66. Sundström, L., Brolin, A., and Hartler, N. (1993). “Fibrillation and its importance for the properties of mechanical pulp fiber sheets,” Nordic Pulp & Paper Research Journal 8(4), 379-383. DOI: 10.3183/npprj-1993-08-04-p379-383 Ushakov, A. V., Alashkevich, Yu. D., Kozhukhov, V. A., and Kovalev, V. I. (2020). “Current state and prospects for improving the process of milling fibrous semifinished high concentration processes (review),” Khimiya Rastitel'nogo Syr'ya 2020(4), 315-329. (in Russ.). DOI: 10.14258/jcprm.2020048251 Wathén, R. (2006). Studies on Fiber Strength and its Effect on Paper Properties, Ph.D. Dissertation, Doctor of Science in Technology, Helsinki University of Technology, Espoo, Finland. Article submitted: July 8, 2023; Peer review completed: July 29, 2023; Revised version received: October 3, 2023; Published: October 12, 2023. DOI: 10.15376/biores.18.4.8090-8103 PEER-REVIEWED ARTICLE bioresources.com Ushakov et al. (2023). “Pulp properties & paper,” BioResources 18(4), 8090-8103. 8102 SUPPLEMENTARY APPENDIX Table S1. Observations № Х1 Х2 Х3 Х4 Y1 Y2 Y3 Y4 Y5 Y6 Y7 Y8 Y9 Y10 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 2500 2500 2500 2500 2000 2000 2000 2000 2500 2500 2500 2500 2000 2000 2000 2000 2500 2500 2500 2500 2000 2000 2000 2000 2500 2500 2500 2500 2000 2000 2000 2000 2500 2500 2500 2500 2000 2000 2000 2000 2250 2250 2250 2250 2250 2250 2250 2250 2250 2250 2250 2250 2250 2250 2250 2250 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.5 1.5 1.5 1.5 0.5 0.5 0.5 0.5 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 20 20 20 20 20 29 20 20 20 20 20 20 20 29 20 20 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 20 20 20 20 10 10 10 10 15 32 47 62 15 27 48 62 15 30 42 62 15 31 40 61 15 28 43 60 15 31 48 62 15 29 45 58 15 33 46 60 15 30 45 63 15 29 43 61 15 30 46 57 15 27 43 63 15 30 45 57 15 30 45 60 0.897 0.837 0.837 0.819 0.897 0.854 0.848 0.805 0.897 0.824 0.821 0.810 0.897 0.813 0.801 0.798 0.897 0.849 0.847 0.832 0.897 0.814 0.805 0.779 0.897 0.844 0.85 0.826 0.897 0.837 0.801 0.792 0.897 0.904 0.844 0.816 0.897 0.863 0.852 0.833 0.897 0.849 0.828 0.811 0.897 0.861 0.83 0.839 0.897 0.727 0.698 0.694 0.897 0.863 0.837 0.825 0.846 1.209 1.264 1.431 0.846 0.918 1.048 1.369 0.846 1.029 1.230 1.374 0.846 1.314 1.672 1.822 0.846 1.005 1.027 1.313 0.846 1.342 1.741 2.031 0.846 0.940 1.113 1.342 0.846 1.242 1.544 1.952 0.846 1.028 1.114 1.307 0.846 1.005 1.105 1.178 0.846 0.964 1.171 1.339 0.846 0.952 1.200 1.454 0.846 1.059 1.26 1,36 0.846 1.027 1.336 1.505 0.113 0.139 0.187 0.171 0.113 0.194 0.195 0.218 0.113 0.151 0.151 0.207 0.113 0.087 0.092 0.092 0.113 0.113 0.227 0.191 0.113 0.131 0.117 0.129 0.113 0.145 0.146 0.179 0.113 0.083 0.112 0.126 0.113 0.152 0.182 0.253 0.113 0.155 0.166 0.174 0.113 0.179 0.167 0.174 0.113 0.141 0.154 0.140 0.113 0.119 0.143 0.164 0.113 0.159 0.125 0.121 20.3 20.8 21 21.1 20.3 20.6 20.9 21.3 20.3 21.1 21.4 21.2 20.3 21.2 21.3 21.5 20.3 20.9 20.7 21.3 20.3 20.5 21.4 22.1 20.3 20.8 20.8 20.9 20.3 20.6 20.8 21.6 20.3 20.8 21.2 21.1 20.3 21 21.1 21.3 20.3 20.8 20.9 21.1 20.3 21 21.1 21 20.3 21.1 21.2 21.4 20.3 21.1 21.2 21.2 147 272 302 325 147 262 316 326 147 274 318 355 147 250 299 311 147 275 295 320 147 252 308 313 147 262 294 315 147 227 336 330 147 282 302 371 147 293 297 313 147 293 323 332 147 248 312 315 147 274 315 374 147 275 297 303 134.96 130.74 133.00 134.08 134.96 133.03 134.40 134.89 134.96 131.98 132.82 134.94 134.96 125.38 125.03 124.32 134.96 135.34 135.04 136.34 134.96 128.41 128.37 128.73 134.96 135.14 136.48 135.82 134.96 129.20 131.40 129.05 134.96 133.15 135.31 136.06 134.96 132.29 134.10 135.41 134.96 132.85 134.07 133.86 134.96 132.95 135.15 135.37 134.96 133.27 134.25 134.76 134.96 135.40 136.49 135.89 4 6 18 47 4 7 28 31 4 12 24 88 4 13 70 150 4 5 16 41 4 16 95 262 4 7 18 41 4 18 98 286 4 8 18 65 4 21 51 98 4 10 35 65 4 8 19 96 4 9 22 41 4 14 41 54 998 2592 3950 5102 998 2921 4526 5184 998 3620 4446 5678 998 3573 6172 6460 998 2386 3248 4567 998 4037 5225 7036 998 3085 3785 5555 998 3538 4372 5925 998 3086 4372 6501 998 3538 5061 5678 998 3703 4773 5802 998 3332 5020 5473 998 3332 3950 5020 998 3456 4899 6131 55 112 155 212 55 102 151 184 55 115 154 200 55 138 209 210 55 108 133 208 55 140 221 277 55 112 154 187 55 143 192 242 55 118 163 214 55 121 160 197 55 122 161 201 55 124 161 203 55 122 138 186 55 117 170 218 216 340 471 549 216 262 471 523 216 314 471 601 216 497 601 627 216 313 392 601 216 497 523 601 216 392 418 470 216 523 575 627 216 392 496 471 216 366 471 497 216 471 497 523 216 366 471 602 216 313 471 549 216 392 418 523 PEER-REVIEWED ARTICLE bioresources.com Ushakov et al. (2023). “Pulp properties & paper,” BioResources 18(4), 8090-8103. 8103 Table S2. Statistical Characteristics Observations Statistics Y1 Y2 Y3 Y4 Y5 Y6 Y7 Y8 Y9 Y10 Count 56 56 56 56 56 56 56 56 56 56 Average 0.840 1.16 0.14 20.88 262.77 133.46 38.0 3644.3 137.94 410.21 Standard Deviation 0.047 0.29 0.03 0.42 72.69 2.89 56.06 1843.48 60.54 138.13 Minimum 0.694 0.84 0.08 20.3 147.0 124.32 4.0 998.0 277.0 216.0 Maximum 0.904 2.01 0.25 22.1 374.0 136.48 286.0 7036.0 222.0 627.0 Coeff. of Var. (%) 5.66 25.5 26.0 2.04 27.66 2.16 147.5 50.58 43.89 33.91