Mid Sweden University

Increased knowledge on the correlation between pulp processing, fibre-properties and paper properties is required to improve fibre-based products. Part 1 of this investigation deals with the effects of HC and LC refining on fibre properties development. LC refining reduced curl and increased tensile index in a manner similar to hot disintegration whereas HC refining increased curl slightly. In this second part, the correlation between fibre curl and handsheet properties of thermomechanical pulp, subjected to low consistency (LC) refining and hot/cold disintegration is examined. Fibre curl decreased by laboratory disintegration and LC refining and showed a linear correlation with increased tensile index and tensile stiffness. Evaluation of fibre property distributions gave a more detailed description of the development of fibre properties. These revealed that disintegration and LC refining gave different fibre curl versus fibre length distributions, even when their average values were similar. These results confirm that analysing fibre property distributions contributes to a more detailed knowledge of the development of pulp quality. Hot disintegration before laboratory testing exaggerated pulp quality and increase internal fibrillation and can therefore be questioned. When hot disintegration is performed before pulp analyses, the impact of LC refining on paper properties may be misjudged. 

We present a method for pulp particle characterization based on a truncated lognormal mixture (TLM) model, as motivated by size statistics of organisms. We use an optical fiber analyzer to measure the length–width distribution of kraft-cooked roundwood or sawmill sources, of chemi-thermomechanical pulp (CTMP) samples from roundwood or sawmill sources, and the same CTMP samples after kraft post-processing. Our results show that bimodal TLMs capture salient features of the investigated pulp particle distributions, by decomposition into a large-particle and a small-particle fraction. However, we find that fibers from sawmill sources, which have not undergone mechanical treatment, cannot be described by TLM, likely due to non-random sampling. Within the confines of our dataset, the TLM characterization predicts laboratory sheet properties more effectively than conventional averaging methods for pulp particle size distributions. The TLM characterization is intended as a tool for controlling the pulp production process towards higher product quality, uniformity, and energy efficiency, pending further mill trials for validation. 

Particles in mechanical pulp show a wide variety but are commonly described using averages and/or collective properties. The authors suggest using distributions of a common bonding factor, BIND (Bonding INDicator), for each particle. The BIND-distribution is based on factor analysis of particle diameter, wall thickness, and external fibrillation of several mechanical pulps measured in an optical analyser. A characteristic BIND-distribution is set in the primary refiner, depending on both wood and process conditions, and remains almost intact along the process. Double-disc refiners gave flatter distributions and lower amounts of fibres with extreme values than single-disc refiners. More refining increased the differences between fibres with low and high BIND. Hence, it is more difficult to develop fibres with lower BIND. Examples are given of how BIND-distributions may be used to assess energy efficiency, fractionation efficiency, and influence of raw material. Mill scale operations were studied for printing-grade thermomechanical pulp (TMP), and board-grade chemi-thermomechanical pulp (CTMP), both from spruce.

Particles in mechanical pulp are a heterogeneous popu-lation, but commonly described using averages based on wide and skewed distributions. It was found that these aver-ages may lead to erroneous conclusions regarding the char-acter of the material and also how the material has been de-veloped along the process. This study is based on measure-ments of individual particle dimensions (length, curl, and ex-ternal fibrillation) in mill operation of CTMP and TMP as detected in an optical analyser.

For a more profound understanding of how a process works, it is essential to have a relevant description of the material being processed. With this description, it will be easier to evaluate and control processes to produce more uniform products with the right properties. The focus of this thesis is on how to describe mechanical pulps in ways that reflect its character.

Mechanical pulps are made from wood, a highly heterogeneous material. Common practice within the pulping industry and academy is to describe mechanical pulps and its wide variety of particles in terms of averages. The energy efficiency of the mechanical pulping process is usually calculated without taking into account the characteristics of the wood fed to the process. The main objective of the thesis is to explore ways to make more detailed descriptions of mechanical pulps. A second objective is to propose useful ways to visualise these descriptions.

The studies were carried out in full-scale mill operations for TMP of publication grades and CTMP for board grades with Norwegian spruce as raw material. The particles in the pulps were analysed in an optical particle analyser for several properties such as length, curl, wall thickness, diameter,and external fibrillation for 10,000 to 60,000 particles per sample to cover their wide property variation. The data was analysed by factor analysis, a method to reduce the multidimensional data space, and also compared with data simulations.

Several examples were identified where averages based on wide and skewed distributions may hide useful information and therefore result in misleading conclusions regarding the fibrous material and process performance. A method was developed to calculate the distribution of a common bonding factor, BIND (bonding indicator) for individual particles. This factor is calculated from external fibrillation, wall thickness and diameter measured in an optical particle analyser. Distributions of BIND is one way to characterize and visualise the heterogeneity of mechanical pulp. A characteristic BIND-distribution is set in the primary refiner stage, depending on both wood and process conditions and remains mostly intact through the process.

It was demonstrated that both BIND-distributions and 4D maps of the measured property distributions could be used to assess the tails of the distributions (extreme values), energy efficiency, and fractionation efficiency in a new way. It was even possible to get a measure for energy efficiency for a primary stage refiner, since a method was developed where the wood raw material was evaluated in the same way as the pulp discharged from the refiner.

It was demonstrated that the average length-length-weighted fibre length, commonly referred to as the average weight-weighted fibre length, is a relevant way to express the amount of long fibres, i.e. “length factor”. The commonly used average length-weighted fibre length may lead to erroneous conclusions. Through data simulations of curl and fibre length on particle level it was found that today’s analysers may underestimate the true length of the particles, especially if they are prone to be curled. As a result, theranking of pulps may be altered.

It was concluded that although there is an ISO standard, or long-time used property, it does not necessarily imply that it is a relevant method. Misleading conclusions may be drawn based on current methods; here, modifications of these methods are suggested.

The main contribution of this study is the finding that that a highly heterogeneous material such as mechanical pulps could be described in new ways through visualisation of data in 4D maps. These maps reveal casualconnections and more pertinent questions may be raised in thecommunication along the chain product-pulp-wood.

Going beyond averages may reveal discrepancies in the process and material that were previously unknown, and lead to a more profound understanding. It seems that the mechanical pulping process can be even further simplified than previously expected. It has been concluded that to operate the process more efficiently, and for make products with just the right quality, the main focus should be on the raw material and the primary refiner stage from a heterogeneity point of view.

För att få en fördjupad förståelse av hur en process fungerar är det väsentligt att ha en relevant beskrivning av materialet som processas. Det gör det enklare att utvärdera och styra processer för att tillverka produkter med jämn kvalitet med just de rätta egenskaperna. Fokuset i denna avhandling är hur mekaniska massor kan beskrivas för att reflektera deras karaktär.

Mekaniska massor är gjorda av ved som är ett mycket heterogent material. Inom massaindustrin och den akademiska världen beskrivs nästan uteslutande mekaniska massor och dess stora variation av partiklar i form av medelvärden. Energieffektivitet hos processer för mekaniska massor beräknas vanligtvis utan att ta hänsyn till hur vedråvaran är beskaffad. Huvudmålet med denna avhandling var att undersöka sätt att ge en mer detaljerad beskrivning av mekaniska massor. Ett andra mål var att föreslå användbara sätt att visualisera beskrivningarna.

Undersökningarna genomfördes i fabriksskala för termomekanisk massa (TMP) för tryckpapper respektive kemitermomekansik massa (CTMP) förkartong, alla med gran som råvara. Partiklarna som utgör massa analyserades in en optisk partikelanalysator som mätte egenskaper såsom längd, curl (krokighet), väggtjocklek, diameter samt extern fibrillering. För att täcka den stora variationen i partiklarnas egenskaper mättes för varje prov 10 000 till 60 000 partiklar. Data analyserades med faktoranalys, en matematisk metod at kondensera datarymden. Dessutom modellerades samband mellan fiberegenskaper med datasimulering.

Åtskilliga exempel visades där medelvärden baserade på breda och skeva fördelningar kan dölja användbar information och därmed ge missvisande slutsatser vad avser såväl fibermaterialet som processen. En metod utvecklades för att beräkna distributionen av en så kallad gemensam bindningsfaktor, BIND (bonding indicator) på partikelnivå. Denna faktor är beräknad ifrån extern fibrillering, väggtjocklek samt diameter mätta in en optisk partikelanalysator. Fördelningen av BIND är ett sätt att beskriva och visualisera heterogeniteten hos mekaniska massor. En karaktäristisk BIND fördelning skapas i det först raffineringssteget, beroende på både veden och processbetingelserna, och bibehålls nästan intakt i de följande processtegen.

Det visades att både BIND-fördelningarna samt fyrdimensionella kartor baserat på rådata av partikelegenskaperna kunde användas för att beskriva ”svansar” i distributionerna (extremvärden). Dessutom är det möjligt att på ett nytt sätt få mått på energieffektivitet samt fraktioneringseffektivitet. Till och med energieffektivitet hos ett primärraffineringssteg var möjligt att få ett mått på eftersom vedråvaran utvärderades på samma sätt som massan som kom ut från nämnda processteg.

Vidare visades det att längd-längd-viktat medelvärde av fiberlängd, även benämnt vikt-viktat medelvärde, är ett relevant mått för att uttrycka andelen långfiber i en massa. Det vanligtvis inom branschen använda längd-viktademedelvärdet av fiberlängd kan leda till felaktiga slutsatser. Genom datasimulering av curl och fiberlängd på partikelnivå framkom det att dagens fiberanalysatorer kan undervärdera den verkliga längden hos partiklar särskilt om de är curlade. Rangordningen av massor kan bli omkastad.

Trots att det finns ISO standarder eller under lång tid använda egenskaper innebär det inte att det är en relevant metod utan kan leda till att felaktiga slutsatser dras. Modifiering av existerande metoder samt nya metoder har föreslagits.

En viktig slutsats i denna avhandling är att ett mycket heterogent material, såsom mekaniska massor, är möjliga att beskriva genom visualisering av data i form av fyrdimensionella kartor. Dessa visar på orsakssamband samt underlättar kommunikationen i kedjan produkt-massa-ved. Mer relevanta frågor är därmed möjliga att ställa.

Genom att gå bortom medelvärden är det möjligt att upptäcka avvikelser i processen och materialet, icke kända tidigare, och ge en mer grundläggande förståelse. Det finns starka indikationer på att den mekaniska massaprocessen är möjlig att förenkla mer än tidigare antagits. Dessutom kan slutsatser dras hur processen kan köras effektivare och göra det möjligt att tillverka produkter med rätt kvalitet genom att fokusera på vedråvaran samt primärraffineringssteget, allt utifrån från ett heterogenitetsperspektiv.

Production of mechanical pulps requires high specific electrical energy compared to many other attrition processes. In Scandinavia, the lowest specific refining energy for production of thermomechanical pulp is around 1800 kWh/t for newsprint quality, which is roughly 60 times higher than for crushing of stone to a similar size distribution. The high specific energy demand for refining has naturally motivated large efforts in the search for improved efficiency. It is always practical to be able to quantify improvements in efficiency for comparison of process designs and of different machine types. However, there is no commonly accepted definition of efficiency for mechanical pulping processes. In published work within mechanical pulping, energy efficiency has been presented in different ways. In this paper, we discuss definitions of energy efficiency and aspects that ought to be considered when energy efficiency is presented. Although focus of this work is on energy efficiency for refiner processes, the principles can be applied to other types of mechanical pulping processes such as stone groundwood. 

The aim of this study was to evaluate changes in fibre properties with high (HC)- and low consistency (LC) refining of TMP and determine how these contribute to tensile index. Two process configurations, one with only HC refining and another with HC refining followed by LC refining were evaluated in three TMP mainline processes in two mills using Norway spruce. An increase in tensile index for a given applied specific energy was similar for all LC refiners in the three lines, despite differences in the fibre property profiles of the feed pulps. Compared with only HC refined pulps at a given tensile index, HC+LC refined pulps had greater fibre wall thickness, similar fibre length, strain at break and freeness, but lower light scattering coefficient, fibre curl and external fibrillation. The degree of internal fibrillation, determined by Simons' stain measurements, was similar for both configurations at a given tensile index. The results indicate that the increase in tensile index in LC refining is largely influenced by a decrease in fibre curl and in HC refining by peeling of the fibre walls. Compared at a given tensile index, the shive content (Somerville mass fraction) was similar for both HC+LC and HC refining. 

Heterogeneity may be a most proper word to describe paper and board. Wood, the raw material to produce mechanical pulps, is almost as heterogeneous a material as the products. Looking in a microscope at a mechanical pulp it is obvious that it consists of a huge amount of small particles, which vary very much in several aspects: length, curl, width, wall thickness, fibrillation etc. In spite of that, we (scientists, mill employees, suppliers, consultants, researchers) commonly describe a pulp in terms of collective properties such as averages of fibre dimensions, hand sheet properties, dewatering ability etc. In other words: we describe mechanical pulp as a continuum. This may be more common than we believe. A review of the variables used in the more than 5,000 graphs in the preprints of the 23 International Mechanical Pulping Conferences (IMPC) between 1973 and 2018 showed that 97% of the variables described mechanical pulp as a continuum. Hence, only 3% of the variables reflected the heterogeneous nature of mechanical pulps.   Many authors point out the benefits of describing the character of a material by as few and as independent properties as possible. In 1957, Steenberg stated: “Any valuable theory must be supposed to include a number of independent factors”. However, within the mechanical pulping area it is common practise to evaluate the character of pulp with respect to a wide range of more or less correlated measured pulp and fibre properties. In the review of the last 23 IMPCs we found that more than 96% of the variables examined in the articles where of that kind. In our efforts to describe a mechanical pulp with respect to its highly heterogeneous nature we have among others been inspired by Forgacs who worked with independent common factors in order to characterize mechanical pulps. He presented a paper in 1963 where he states that at least two independent factors are needed to describe a mechanical pulp. They reflect shape and length of the particles. Also Strand has inspired us with his work in the 80’s where he used factor analysis on a huge database to derive two independent common factors reflecting “bonding” (Factor 1) and “fibre length” (Factor 2). Examining a few independent common factors instead of several conventionally measured properties, which are more or less correlated, makes it easier to get an overview of the status of the processed material. Therefore, it is a little surprising that independent common factors constitute only 4% of the variables presented at the IMPCs since 1973. Strands approach was tested in a long-term evaluation in two mills. It was possible to produce a paper product more uniform in quality, compared to common practise. However, none of the above-mentioned independent factors reflects the heterogeneity of the pulp.   Paper and pulp makers commonly agree that, within certain limits, uniformity is the most important characteristic of both the pulp and the paper. If we know how to perform uniformly, we may also be able to move into other operating areas (or volumes) in a controlled way. However, there is no common agreement on how to define "uniformity". Papermakers are still to a great extent specifying their demands on the pulp in terms of dewatering ability and average length-weighted fibre length although the correlation to product quality is vague and weak and varies over time. Almost since the advent of mechanical pulping processes, the operators have for process control had readings of dewatering ability of a pad consisting of billions of particles expressed as mL of water and average length-weighted fibre length of the pulp, which are far from being independent factors. Variations in any of these two properties may depend on variations in a combination of several more underlying factors. Therefore, it is hard to know what actions the operators should take to avoid running off spec. So far, the main development in the concept with dewatering and length to assist the operators have been firstly, to get time trends of these variables on a DCS screen instead of on a piece of paper in the control room, and secondly to get the readings more frequently with on-line analysers. During the same period, there have been an immense development of refiner concepts, fractionation, process design, modelling, use of raw material, fibre characterization, and new products.   By putting more attention to reality and describing mechanical pulp as a heterogeneous material, which the mechanical pulps truly are, we hope to be able to get a more profound understanding how wood particles are developed along the process all the way to product. We also hope to give the operators in the mechanical pulping plant a more realistic description of the material they are supposed to deliver to the paper and board makers in order to facilitate their possibilities to produce a more uniform product quality at minimum cost.  The aim of our presentation is to share some of our insights and reflections how to describe the heterogeneous nature of mechanical pulps to the mill operators.

We have applied factor analysis on particle level based on measurements in an optical analyser of fibre diameter, fibre wall thickness, fibre length, and fibre fibrillation. Examples will be presented of how the raw material and the process have set characteristic fingerprints in terms of the distribution of an independent common bonding factor on particle level. It is fascinating to see how much that may be hidden behind averages, c.f. Rosling et al. (2018) who warn against comparing averages, which often obstruct a more profound understanding of a subject.   In our presentation, we intend also to discuss how a description of the heterogeneity of the material may be used to get measures of energy efficiency of the process, separation efficiency of fractionation equipment, and how to link fibre characteristics to properties of products. Some reflections will also be shared on what we think is further required to get a more realistic description of the heterogeneous material we call mechanical pulp. Reference:  Rosling H., Rosling Rönnlund A., Rosling O., (2018), ”Factfulness”, Natur & Kultur,  ISBN 978-91-27-14994-6

The purpose of this study was to investigate the applicability of different ways of calculating the average fibre length based on length measurements of individual particles of mechanical pulps. We have found that the commonly used average length-weighted fibre length, which is based on the assumption that coarseness is constant for all particles, as well as the arithmetic average, may lead to erroneous conclusions in real life as well as in simulations when used as a measure of the amount of long fibres. The average length-length-weighted fibre length or a weighting close to that, which to a larger extent suppresses the influence of shorter particles, is a relevant parameter of the "length" factor, i. e. amount of long fibres. Our findings are based on three studies: refining of different assortments of wood raw material in a mill; data from LC refining in mill of TMP, including Bauer McNett fractionation; mixing of pulps with different fibre length distributions. If the acceptable average fibre length for different products can be lowered, the possibility of reducing the specific energy input in refining will increase. Therefore, we need a reliable and appropriate way to assess the "length" factor.

Paper and wood are highly inhomogeneous materials. When describing the mechanical pulp itself, we allcommonly ignore that it is an inhomogeneous material. We have realized that just a very small fraction of stifffibres are enough to impair the printability of the product. In this paper we share some of our reflections andattempts how to describe the inhomogeneous nature of mechanical pulps. A method denoted BIN is underdevelopment based on independent common factors and paying attention to the inhomogeneity of the material.The method may give the possibility to describe the nature of TMP/CTMP/SGW in a more relevant way comparedto todays practice. Hence the paper and board makers may be able to deliver more uniform products at “goodenough” level at lower costs. We have realized that because a method or opinion is well spread (sometimes usedby almost everybody) it does not necessarily mean that it is relevant. A couple of myths have been reflected uponand in our opinion they remain just myths. By putting more attention to reality and describing mechanical pulp asan inhomogeneous material we hope to be able to rid ourselves and the mechanical pulping community of someother myths circulating (some still to be discovered).