Energy efficiency in mechanical pulping – definitions and considerations

Efficiency

The word efficiency originates from the 16^th century and is derived from the Latin word efficere = accomplish. The dictionary definition of efficiency is: The ratio of the useful work performed by a machine or in a process to the total energy expended or heat taken in. Lexico (2021)

where E U is the useful work and E S is the total energy supplied.

In many applications, the useful work has the same unit as the supplied energy, e. g. for a combustion engine the useful work and the supplied fuel can both be expressed in joules and thus the efficiency is a unitless share that can be expressed in percent: “The efficiency of this diesel engine is 35 %”.

Figure 1

Symbols used in block diagrams.

Figure 2

A conventional TMP process with three system borders shown.

System definition

Energy efficiency can be evaluated for a whole process or for a single unit operation such as a refiner. It can also be defined for a section of a process such as the whole reject treatment system in a pulp line including screening and reject refining. In Figure 1, symbols used in block diagrams are explained. In Figure 2, three examples of system boundaries are shown for a TMP process. It is important to specify the system borders when efficiency is discussed.

Useful work

The numerator in Equation 1 represents the useful work produced by the system. For mechanical pulping, it has been suggested that the term “useful work” could be expressed as the theoretically minimum energy needed to create all surfaces in a certain pulp. Over the years, different approaches have been used to estimate this theoretical specific energy resulting in a wide range of the numbers. Since the wood raw material and the refining process are highly heterogeneous, it is a challenging task to determine a theoretical minimum SRE. The complex mechanisms in refining and different theoretical calculations have been comprehensively described by Berg (2001) and Kerekes (2010). Thus, relating the “useful work” to a theoretical minimum specific energy is not a practical definition for mechanical pulping.

Even though Campbell (1934) discarded a definition for efficiency based on pulp quality, it is worth discussing this subject. Atack (1981) pointed out that specific energy must be related to pulp quality. If efficiency is defined as the ratio of a measured fibre or pulp property to the supplied energy, then the result cannot be expressed as a share in percent. Since the pulp quality out from a process depends, not only on the process conditions but also on the feed properties (of wood chips or pulp), the numerator in Equation 1 must be the change in a fibre or pulp property over the studied process section. Immediately the question arises: which pulp property or properties should then be used? Normally, pulp quality is represented by means of a relatively large number of pulp properties such as; freeness, tensile index, tear index, light scattering coefficient, fibre length, etc. However, Jones (1962) showed that many of these properties are interrelated. Based on such findings, Forgacs (1963) suggested that it should be possible to describe pulp quality with a small number of independent basic properties. Strand (1987) as well as Ferritsius and Ferritsius (1997) have showed that pulps can be characterised with two independent factors reflecting “bonding” (Factor 1, F1) and “fibre length” (Factor 2, F2) based on a large number of pulp and fibre properties. These factors correlated to Forgacs’ S- and L-factors. Another way would be to relate the efficiency to a change in a fibre property distribution. As a comparison, for crushing of stone, the specific energy is often related to the size distribution of the produced material (Ciężkowski et al. 2017). The use of pulp quality to express the “useful work” to will be discussed further below.

The recovered steam from HC refiners has, in most cases, a value and could be considered as part of the “useful work” produced by the process. It is of course not mechanical work (even though this steam does contain exergy), but it still has a value, and changes in the steam production that might be a result of improved refining efficiency should be considered. One example is replacing HC refiners with low consistency (LC) refiners. Another example is that double disc (DD) refiners in some cases need high pressure feed-steam (10–12 bar) which should be considered. These aspects are mill-specific, therefore we cannot make general statements.

Supplied energy

The widely used expression energy consumption is of course not strictly correct. Energy cannot be consumed only converted from one form to another. In mechanical pulping, the major part of the supplied energy is electrical energy utilized in refiners or grinders, which can be “consumed” and transferred to heat.

The denominator in Equation 1 is the “total expended energy”. In mechanical pulping the supplied energy is usually related to the amount of produced pulp, which is called specific energy. Energy is expressed as joules or, more often for mechanical pulping, as watthours or horsepowerdays and the amount of pulp is usually measured in metric tons (1000 kg). Thus, the specific energy can be derived by dividing the motor power of a refiner with the production rate and it is usually reported in kWh/t. Many mills present the production rate at 10 % moisture content, which is referred to as “air-dry” i. e. the reported specific energy is in kWh/adt. For “dry tonnes” the abbreviation bdt (“bone-dry” ton) is sometimes used. Some published work, do not specify the units for the production rates used (e. g. Nurminen 1999, Kure et al. 2000). Presumably, the production rates were presented in dry tons, but it is preferable to specify. The production rate can be more difficult to measure than the refiner motor power, but how it was measured should be stated.

Normally, the reported specific energy in published work is for the refiners only. Here, we denote the specific refining energy ‘SRE’. Refiners are powered by electrical motors. The losses in bearings, sealings and the motor itself are usually included in the SRE. For large refiner motors, the efficiency is usually around 98 % (ABB 2021). For LC refiners it is quite common to study the pulp development as a function of “net” SRE (motor power – no-load power). Some papers do not specify if the SRE reported is net or total (gross) (e. g. Fernando et al. 2013). The net SRE is useful in theoretical studies and from a refiner control perspective. In our opinion, when energy efficiency is evaluated, the gross SRE (including no-load, and other losses) should be used. When a whole process is studied, the total specific energy can be interesting i. e. including all auxiliary energy such as pumps, agitators, presses, screw conveyors, etc. Some process concepts that has been proposed require more equipment and thereby more auxiliary energy which should be considered. An example of such a process is the inter-stage screening (ISS) process (Bousquet et al. 2016), in which a larger number of screens, larger screw press capacity, etc, is required. Even if the additional auxiliary specific energy is not high, it is preferable to report it. As an example, Bousquet et al. (2016) did not report the additional auxiliary energy and the SRE reported was related to an undefined “quality index” and not to standardised pulp properties.

As mentioned above, the “useful work” should be related to the “total energy expended”. The term “total energy” can be expanded to include other applied resources than electrical energy, such as chemicals. This has sometimes been neglected. As an example, Gorski et al. (2012) reported an energy reduction of 450 kWh/t for the so called Advanced TMP (ATMP) process compared to a conventional two-stage TMP process. In that case, 31 kg bisulphite per ton of pulp was added in the refining. In Sweden, the cost for this quantity of bisulphite equates to a specific energy of approx. 500 kWh/t. This means that the energy reduction was actually −50 kWh/t! Of course, adding chemicals in the refining process can affect other properties than the specific energy, such as brightness, which can reduce the need of bleaching chemicals later on in the process. Such effects should be considered when the efficiency of the total process is evaluated.

Some chemical treatments, e. g. ozone and highly alkaline peroxide, reduce the pulp yield (Lecourt et al. 2007, Pan 2004) which also should be taken into account when efficiency is presented.

For an overall mill performance, the total cost for making products at a specified quality is important. Thus, the total amount of spent resources should be considered. This could be referred to as “cost efficiency”.

Efficiency – Pulp property change over a process

As mentioned above, the change in a fibre- or pulp property over a process stage can be one way to express the “useful work”. Thus, the efficiency can be calculated as per Equation 2.

where Δ P is the change in a fibre or pulp property over the process stage and E S is the supplied energy. Figure 4 shows two examples of process sections where this definition of efficiency can be used (Sandberg et al. 2017a).

Figure 4

One way to define energy efficiency is to measure the change in a pulp property for a process section divided by the supplied specific energy. Two process sections are shown where the efficiency can be calculated according to Equation 2 (Sandberg et al. 2017a).

As an example, in Figure 5 tensile index is used to calculate the efficiency as given by Equation 2 for the two process sections shown in Figure 4.

Figure 5

An efficiency based on tensile index, η (Nm/g)/(MWh) as given by Equation 2, for the process sections in Figure 4. Data from Sandberg et al. (2017a).

In the Stora Enso Kvarnsveden mill (Borlänge, Sweden), one of the independent factors mentioned above, F1 (Factor 1, bonding), was applied for process control (Ferritsius and Ferritsius 1997). The change in F1 over a process stage divided by the applied energy was used to get an estimate of the energy efficiency. Another way to handle the heterogeneity of the fibre material is to use the change in the distribution of a given particle property over a process stage to quantify the “useful work” (Ferritsius et al. 2019). This is ongoing work that will be published soon.

In some cases, it can be difficult to use the change in a quality parameter to express the useful work, e. g. when the system feed is wood chips it can be impossible to measure some feed properties, e. g. tensile index or fibrillation of wood chips. Thus, this definition is mainly suitable for process stages were the feed is a pulp that is not “too coarse”.

Efficiency improvement at a given pulp property

As discussed above, it might not be possible to calculate the efficiency (defined by Equation 2) of a certain process if the feed is woodchips. However, for two operating points of the same process or when two processes are compared, a relative difference in efficiency can be calculated, provided that pulp properties are available at the same level and that the feed material is the same for the two cases. This is illustrated for tensile index in Figure 6. The relative difference in efficiency of the two processes/operating points is then:

Where T I x is an arbitrary tensile index of the feed material. If the processes are fed with the same raw material and T I 1 is equal to T I 2 , Equation 3 reduces to:

Figure 6

A relative difference in efficiency can be calculated for two processes or a process operating at two different conditions if the feed material is the same.

Raw material

The wood raw material has a large influence on many pulp and fibre properties and how they develop in refining (Corson 1984, Miles and Karnis 1995, Tyrväinen 1995, Persson et al. 2003). If possible, it is advantageous for the evaluation of a pulping process to characterize the wood raw material more precisely than by only giving the tree species and the share of sawmill chips.

Pulp quality

One pulping process can require considerably lower specific energy than another process to achieve a certain pulp property. However, the outcome might be different if comparing the specific energy required to achieve a different pulp property. This is shown in Figure 7 for a second stage main line refining. The LC refiners require around 50 % lower specific energy than the HC refiners to produce a pulp with a given tensile index, Figure 7A, whereas for a given light scattering the two refiner types require similar specific energy, Figure 7B.

Figure 7

Second stage main line refining of TMP. A, LC refining require lower specific refining energy (SRE) than HC refining for a given tensile index, but for a given light scattering, B, there is no difference. Data from Sandberg et al. (2017a).

When the performance of processes is compared, it should be kept in mind that values of measured fibre and pulp properties can differ considerably if pulps are analysed in different laboratories.

Refining curves

The performance of refiners is often evaluated by making “refining curves”, which are produced by changing one or more of the input variables, e. g. hydraulic pressure (disc gap), dilution water flow or production rate. Refining curves are usually performed to get a range of pulp properties at different operating conditions. In Figure 3, two examples of refining curves are shown for a second stage HC SD main line refiner and a single stage HC DD chip refiner in the Holmen Braviken mill, Norrköping, Sweden. Data for the primary SD refiner (i. e. feed to the second stage refiner) is also shown in Figure 3. For both refiners, the hydraulic pressure (gap) was changed at constant production rate to generate the data. The refiner dilution water flows were adjusted to maintain constant blow line consistencies.

The slope of a refining curve depends on which refiner variables are changed. In Figure 8, data is shown for an RGP68DD refiner (A) (Kvarnsveden mill) and an RTS refiner (B) (Holmen Hallsta mill, Hallstavik, Sweden) for which production rate and disc gap were varied. Changing the production rate gives a lower slope than changing the gap. Hill et al. (2017) state that most refiners generate refining curves (tensile index versus SRE at changed disc gap) with a slope of 20 (Nm/g)/(MWh/adt). Our experience is that this is the case for SD refiners when the disc gap is changed, which also is the case for the second stage SD refiner in Figure 3. However, other refiner types, e. g. DD refiners and RTS refiners, generate considerably higher slopes when the gap is changed, Figures 3, 8A and 8B (RTS is a refiner type (Andritz) operating at higher RPM. RTS is an abbreviation of Retention time, Temperature and Speed). Obviously, it is important to evaluate the effect of more than one variable when refining curves are generated.

Figure 8

Refining curves generated by changing disc gap at two production levels for A: an RGP68DD HC refiner (Stora Enso Kvarnsveden mill) and B: for an SB170 RTS¹ refiner (Holmen Hallsta mill). In A, numbers in the legend show production rate in adt/h and disc gap in mm and in B, numbers show production rate in adt/h and refiner power in MW. Process parameters are summarised in the Appendix.

Comparisons

Gorski et al. (2011) reported an energy reduction of 650 kWh/t, corresponding to 37 %, for an ATMP process compared to a two stage TMP process, Figure 9. The comparison was made at tensile index 30 Nm/g. If the comparison would have been made at a tensile index level more appropriate to a final pulp for newsprint, e. g. 44 Nm/g, the energy reduction would have been 550 kWh/t which corresponds to 22 %.

Figure 9

Refining curves for different process configurations, from Gorski et al. (2011). Comparison of the SRE was made at a relatively low level of tensile index, 30 Nm/g.

Thus, the level chosen for comparison can affect the SRE reduction expressed both in kWh/t and in percent. Making comparisons in percent can be misleading.

Efficiency versus energy saving

A high efficiency for an equipment, as defined by Equation 2, over a limited range in SRE does not necessarily mean that the same efficiency will be attained over a wider range. As an example, LC refiners can give a relatively large increase in tensile index at a low SRE, as shown for the first LC refiner in Figure 10. In this case, the high efficiency can to some extent likely be explained by fibre straightening (Ferritsius et al. 2020). The second stage LC refiner in the figure has a considerably lower efficiency and thereby the same high gain in tensile index as in the first stage cannot be achieved when adding multiple LC refining stages.

Figure 10

Tensile index increase for a two-stage LC refining process in the Hallsta mill. The efficiency ƞ (as defined by Equation 2) for the first stage is more than twice as high as for the second stage.

Interpolation-extrapolation

Sometimes, data are not available at the level that one wants to make the comparison at. In such cases it is common to interpolate data to a given level. When that kind of comparisons are made, it is preferable not to do the interpolation at the far end of the data set as shown in Figure 9 (Gorski et al. 2011). Extrapolation far outside of the available data range, such as by Chang et al. (2016) and Hill et al. (2017) should be avoided also.