NC State
Kerekes, R. J. (2015). "Perspectives on high and low consistency refining in mechanical pulp," BioRes. 10(4), 8795-8811.


Recent developments in low consistency refining in mechanical pulping have raised questions about the differences between low and high consistency refining. This paper, originally presented to the UBC Energy Reduction in Mechanical Pulping Committee on 10 June 2015 at the PACWEST Conference in Whistler, BC, Canada, discusses the author’s perspectives on these issues as well as on mechanical pulping in general.

Full Article

Perspectives on High and Low Consistency Refining in Mechanical Pulping

Richard J. Kerekes

Recent developments in low consistency refining in mechanical pulping have raised questions about the differences between low and high consistency refining. This paper, originally presented to the UBC Energy Reduction in Mechanical Pulping Committee on 10 June 2015 at the PACWEST Conference in Whistler, BC, Canada, discusses the author’s perspectives on these issues as well as on mechanical pulping in general.

Keywords: Mechanical pulping; Refining; Stone groundwood; Energy

Contact information: Pulp and Paper Centre, 2385 E Mall, University of British Columbia, Vancouver, BC Canada V6T 1Z4; Email:


The discovery of mechanical pulping was a landmark event in the history of paper, indeed in the history of the world. The development of stone ground wood (SGW) pulping in Germany in the 1850’s, followed by sulphite pulping about 10 years later, paved the way for the widespread use of wood fibre in paper. This in turn enabled mass print media such as the penny press to displace what was formerly a ‘rich man’s product,” paper made from various non-wood fibres.

An SGW grinder is shown in Fig 1. In essence, logs are pressed against a grindstone. Grits on the grindstone surface pass over the surface of the wood as shown to scale in Fig 2.

Fig. 1. Grinder to produce stone groundwood pulp. See Acknowledgements for figure credits and source information.

Fig. 2. Grinder grits sliding over wood surface (Atack and May 1958)

The grits of a grindstone accomplish two actions: removal of fibres from the wood (comminution) and development of the fibres into a paper-grade pulp. Comminution is accomplished by a rubbing and sliding action, as opposed to grinding, where the main action is due to cutting by the sharp edges of the grits (Atack and May 1958).

Grits in SGW create rolling friction and consequent heat from hysteresis, to the extent that the wood may char just under the surface. Consequently water is added. Once fibre ends are loosened, they turn in the direction of grinder rotation, as shown in Fig 3, and are eventually released very much as intact fibres (Atack 1977). The free pulp is further “developed” in what is called re-grinding, a heterogeneous process that causes fibre length loss.

Fig. 3. Fibres being peeled from wood surface during grinding (Atack 1977)

Mechanical pulp thus produced is shown in Fig 4a, where it is compared to chemical pulp. A key feature of mechanical pulp is clearly evident: about one third to one half of the fibre mass is in the form of short fibres and fines. In contrast, chemical pulping produces long fibres with relatively few fines. Given this nature of mechanical pulp, Forgacs (1963) defined its two key properties as fibre length, L, and external specific surface, S, with the latter commonly measured by the freeness test.

Fig. 4. Mechanical pulp (left), chemical pulp (right)

The 1950’s saw the introduction of refiner mechanical pulping (RMP). This came after the development of well-engineered disc refiners for chemical pulps in the 1930’s. The three main drivers for RMP were: availability of wood as chips rather than whole logs needed for SGW; less intensive labour; and a continuous process as opposed to the batch SGW process. An RMP refiner is shown in Fig 5.

Fig. 5. Disc refiner and typical plate

In operation, wood chips enter the eye of the refiner where the star-bolt and breaker bars impart a rotational motion to them, giving the centrifugal acceleration that drives the pulp radially through the refiner. At the same time, these bars comminute the wood, to an extent that the fibrous material entering the remaining bar section is a coarse pulp. This was demonstrated by Neill and Beath (1963) using specially-made plates having only the star bolt and breaker bar section. From this they concluded that the wood comminution step required no more than about 400 kWh/t of the typical total 1500 kWh/t, in short about 25% of the energy of mechanical pulping. The rest is for fibre development.

Fibre development in RMP occurs by a mechanism similar to that in SGW, but with bar edges rather that grits sliding over fibres, as shown in Fig. 6. Another difference is that fibres are held in place by bar edges and frictional forces in RMP rather than by the wood matrix as in SGW. Restraint of some form is required for fibres to strain rather than accelerate when force is applied upon them.

Fig. 6. Illustration of fibres in gap between crossing bars (Page 1989)

The next major advance in mechanical pulping was the introduction of Thermo-Mechanical Pulping (TMP) in the early 1970’s. A major driver here was reduction of refining energy. It was known that lignin softens at high temperature, and therefore it is reasonable to expect that the middle lamella binding fibres together would rupture more easily. This would reduce the energy required to comminute the wood into fibres and at the same time inflict less damage on fibres. However, to some surprise, TMP consumed more, not less less, energy than RMP or SGW, as shown in Table 1. At the same time TMP also produced longer fibres, fewer shives, and fewer fines.

The reason for the higher energy is evident in Fig. 7. Wood comminution takes place through the softened lignin of the middle lamella. This produces lignin-rich fibres and consequently more energy is required for the subsequent fibre development. Since most energy is consumed in this stage of the process, overall energy increases. In contrast, breakage through the P1 layer shown in Fig. 7 exposes cellulosic components of the fibre wall.

Table 1. Comparison of Pulp Properties

Fig. 7. Illustration of effect of temperature on fracture zones during comminution (FPInnovations)


The large electrical energy required for mechanical pulp production has long been a major issue. It raised an important question: how much energy is theoretically needed? The first estimate of this was by Campbell (1934). In essence, he multiplied the surface free energy of a crystalline solid, 650 ergs/cm2, by the amount of new surface created in SGW, which he measured to be 1.2 m2/g. This gave an astonishing low efficiency of 0.012% for 1600 kWh/t SGW. Later, Atack (1981) estimated new surface to be around 7 m2/g, which gave an efficiency of 0.07%, which is still very small.

Campbell noted that his estimate did not take into account the energy required to produce new surface. In mechanical pulping the new surface is created by straining wood to rupture, a problem in fracture mechanics. In the early 1960’s, two separate groups conducted experiments on fracture mechanics of wood. Lamb (1960, 1962) employed wood peeling, and Atack et al. (1961) used wood cleavage. The latter group also related their findings to Griffiths crack propagation model, modified for ductile materials. In both studies, the energy of fracture was found to be about 105ergs/cm2, giving an efficiency in the range of 10 to 15% for mechanical pulping. This value was verified later by studies of wood cleavage by Ottestam and Salmen (2001). In summary, the efficiency of mechanical pulping based in fracture mechanics to create new surface is about 13%.

Efforts to measure efficiency of mechanical pulping directly were conducted by determining the lowest limit of energy to produce SGW. Using a specially constructed grindstone at low rotational speed, Salmen et al. (1999) produced paper-grade pulp at
700 kWh/t, giving an efficiency of about 50% based on the conventional energy of 1500 kWh/t. Nevertheless, there remains a large gap between theoretical energy of 13% and the lowest practical efficiency of 50%. Why is this the case?

Fibre development consumes most of the energy in mechanical pulping. Some development may take place by chemical softening without new surface creation, and some by internal cracking (Maloney and Paulapuro 1999). However, Karnis (1994) showed that most fibre development takes place by removal of surface layers of fibres, i.e., by a “de-coarsing” effect, as shown in Fig. 8. He determined that about 50% of the fibre wall is peeled away, with about 28% becoming short fibres and about 27% becoming P200 fines. Fibre flexibility is increased due to its fourth power dependence on outer diameter.

Fig. 8. Effect of energy on coarseness reduction of fibres (Karnis 1994)

Surface material removed by solid bodies sliding over one another is a form of wear. The amount of material removed is related to energy expended, as given by the Holm-Archard equation (Rabinowicz 1995):


Here V is the volume of material abraded, and k is a constant reflecting material properties of the abraded material and surface geometry of the abrading solid. FN and FS are normal and shear forces, is the coefficient of friction, s is sliding distance, and E is the energy expended by the moving hard solid.

The energy E to produce volume V is dictated to a considerable extent by constraints on the desired quality of V. Just as energy for wood comminution is constrained by preservation of fibre length, new surface V is constrained by the desire for either “fibrillar” material to enhance bonding or “chunky” material to enhance opacity. The mechanisms to produce these qualities differ and may be described colloquially as “ripping out chunks” or “jiggling loose fibrils”. Both mechanisms are linked to energy in the form of Eq. 1. However, a second parameter is needed to account for the quality constraint. This parameter is called refining intensity.


Energy in mechanical pulping is applied in a cyclic manner and thus may be described as the product of the number of loading cycles (impacts), N , and the intensity of each Impact, Ii.e.

E=N.I (2)

This concept is illustrated in Fig. 9.

Fig. 9. Refining impacts and intensity at a common energy (Kerekes 1990)

Only two of the variables in Eq. 2 are independent. It is common to use only energy E and intensity, I. There are two approaches to quantifying intensity. One is by a “machine intensity” which describes how the energy is distributed among the bar crossings. The other is by a “fibre intensity” which describes the energy imparted to fibres during each bar crossing. These two intensities differ because not all fibres are impacted at each bar crossing.


Chemical pulps have historically been refined at low consistency (3 to 5%). The earliest work on refining intensity took place for chemical pulp refining, and therefore it is useful to discuss it first. Brecht and Siewert (1966) developed an intensity called the Specific Edge Load (SEL), a machine intensity that remains in widespread use to this day. It is a measure of the energy distribution among bar crossings and is calculated by dividing the net power to the refiner by the number of bars on rotor X number of bars on stator X bar length X rotational speed. Although developed empirically, the SEL has been shown to represent the energy expended per bar crossing per bar length (Kerekes and Senger 2006). Typical recommended values for chemical pulps are SEL= 2 J/m for softwood and 0.2 J/m for hardwood (Baker 1995). Several variants of SEL were developed in later years.

Fibre intensities were developed by Danforth (1969), Leider and Nissan (1977), and through a C-factor by Kerekes (1990). A key need for fibre intensities is to account for the probability of fibre impact at a bar crossing, and thereby the amount of fibre in a gap captured from the adjacent groove. This fraction captured may range from 5 to 20% (Heymer 2009). A significant merit of fibre-based intensity is that it provides a common scientific basis for comparison. For example, although SEL values for chemical pulp hardwood and softwood differ by an order of magnitude, these both represent the same level of fibre intensity (Kerekes 2010). This is understandable, as the composition of hardwoods and softwoods are rather similar. The level is about 10 kJ/kg/impact.


Both RMP and TMP are produced at a high consistency (HC) of about 30%. Refining intensity for this consistency is a difficult matter owing to the complex rheology of HC suspensions. These are heterogeneous three-phase mixtures of wet fibres in steam or air. As shown in Figs. 10 a,b they are discontinuous, compressible mixtures. They have an aerodynamic specific surface of about 50 m2/kg, whereas the hydrodynamic specific surface of a pulp fibre is about 1000 m2/kg (Garner and Kerekes 1978).

Fig. 10. HC Pulp in bulk form (a), showing aerodynamic specific surface (b) (Kerekes et al. 1985)

Given the discontinuous nature of HC pulp, the motive power to force HC pulp to flow through a refiner must come from the refiner itself, specifically from the centrifugal forces imposed by the rotor. In contrast, LC pulp can be pumped through a refiner by an upstream pump. Once in flow, HC pulp tends to be quite heterogeneous, as shown in the discharge flow of a TMP refiner in Fig. 11.

Fig. 11. HC pulp (36% consistency) discharging from a TMP refiner (Kerekes et al. 1985)

The compressibility of HC suspensions means that the amount of pulp in a refiner is unknown. Accordingly, although machine intensity (SEL) may be readily calculated, its meaning with respect to action on pulp is less certain than for LC refiners, which are entirely full of pulp.

The heterogeneity of HC suspensions produces heterogeneity of fibre distribution on bars. This is illustrated by measurements of forces at a point on a bar in HC and LC refiners shown in Fig 12. If one assumes that the temporal variability shown reflects the spatial variability along a bar, bar coverage of HC is much more heterogeneous than is the case for LC.

Fig. 12a. Force measurements on a bar in an HC chip refiner (vertical lines indicate bar crossings (Olender et al. 2007); b: Force measurements on a bar in an LC mechanical pulp refiner (Prairie et al. 2008)


Estimates of refining intensity in mechanical pulping have focused on fibre intensities. An early indirect measure of this in SGW was made by Campbell (Atack 1977). He noted that fibres experienced about 3000 grit impacts. For a specific energy of 1500 kWh/t, this would give an intensity of about 1.6 kJ/kg/impact.

Miles and May (1990) carried out the first rigorous analysis of fibre-based intensity in HC mechanical pulp refiners. They did so by a radial force balance on pulp consisting of an outward centrifugal force, inward frictional resistance, and positive or negative force from steam flow before and after the stagnation point. Doing so gave an average radial velocity of the pulp. From this and the length of the radius, they determined pulp residence time in the refiner. Assuming fibres in grooves were impacted at each crossing, from the number of bar crossings at a point in the refiner and the specific energy they obtained a specific energy per impact, I, to be in the range 0.3 to 0.9 kJ/kg/impact.


Historically, LC refiners for mechanical pulp were used only as post-refiners in paper mills. The principal aim was shive reduction and freeness control. These refiners were found to cut fibres and therefore were deemed unsuitable for fibre development. Although their intensities were not measured, they probably operated in the chemical pulp range, i.e. near to 10 kJ/kg/impact.

At some point it was found that significantly lower intensity could produce fibre development and minimize cutting. This was accomplished by spreading input power over more bar crossings. One way of achieving this was by a refiner called the Beloit Multi-Disc refiner, shown in Fig 13. In it the flow splits into 4 or more working zones, giving a machine intensity in the range of SEL= 0.10 to 0.34 J/m.

Fig. 13. A “Multi-Disc” LC refiner (courtesy GL&V)

An alternate approach to attaining low intensity in recent years has been by use of very narrow bars that permit more bars per plate. Such plates were developed for hardwood chemical pulps by the FINEBARTM and are shown in Fig. 14 in comparison to a traditional cast plate.