Impact of the Drying Temperature and Grinding Technique on Biomass Grindability

: The process of biomass compaction depends on many factors, related to material and process. One of the most important is the proper fragmentation of the raw material. In most cases, more fragmented raw material makes it easier to achieve the desired quality parameters of pellets or briquettes. While the chipping of biomass prefers moist materials, for grinding, the material needs to be dried. As drying temperature changes the properties of the material, these may a ﬀ ect the grinding process. The aim of this work was to determine the inﬂuence of the drying temperature of biomass raw material in the range of 60–140 ◦ C on the biomass grindability. To only determine this e ﬀ ect, without the inﬂuence of moisture, grinding was carried out on the material in a dry state. The research was carried out on a mill with a knife and hammer grinding system, which is the most popular in the fragmentation of biomass. The analysis of particle size distribution and bulk density of the obtained material was carried out. The energy demand for the grinding process was determined and it was shown that drying temperature, grinding system, and mainly type of biomass a ﬀ ects the grindability. Data


Introduction
Biomass, regardless of origin and source (woody, herbaceous, fruit or water) [1], from a chemical point of view, consists of three main components: cellulose, hemicellulose, and lignin. Thus, this kind of biomass is called lignocellulose biomass [2,3], and contains about 50% of carbon and about 6-8% hydrogen [4][5][6]. This is the main reason why it is mainly used as a biofuel [7]. From the point of view of the energy use of biomass, most of the energy accumulated in a given biomass is found in cellulose, hemicellulose, and lignin. Separately, the heat of combustion for cellulose and hemicellulose are similar: 17.6 and 17.9 MJ·kg −1 [4], respectively. Lignin heat of combustion is higher and depends on the type of biomass: lignins from spruce and beech, but also from olive pomace, walnut, and hazelnut shells, have 27-28 MJ·kg −1 [8,9]. Therefore, the biomass heat of combustion is determined by the proportion of lignin in relation to holocellulose, however, despite the different levels of lignin, cellulose, and hemicellulose in biomass [10][11][12][13], the biomass heat of combustion has a narrow range of 18-22 MJ·kg −1 [14,15].
This energy homogeneity means that biomass from a wide range of sources is taken into account as raw material for the production of solid biofuels. These include coniferous and deciduous woods [16][17][18], different species of grasses [19][20][21], and dicotyledonous perennials [22][23][24] as well as straw. Recently, more popular are shells, stones, husks, pomaces [25][26][27], and aquatic biomass [28,29]. distribution (PSD). The biomass for the production of solid biofuels, it is tested based on the EN ISO 17827 standard [69]. Commonly, it is presented in the form of the percentage share of each dimensional fraction in the whole bed or in the form of the cumulative distribution curve of particle size, the sum of dimensional fractions remaining on subsequent sieves. The second method allows for a simple comparison of the ground materials and an indication of which of them has a higher share of fine fractions, and thus which of them have been grounded more. To complete the information about particle size distribution, the median value of grain size d 50 is calculated and defined as the sample particle size, which divides the cumulative distribution curve into two equal parts.
Particle size distribution depends on the origin of the raw material [48,70,71] as well as on the grinding technique [44,72]. In their research, Kirsten et al. ground hay on a cutting, impact, and hammer mill equipped with a 4 mm diameter sieve. In addition, the mills worked with and without an aspiration system to improve material flow through the mill. In each grinding variant, the produced mixture had a different PSD [72]. Puig-Arnavat et al. ground six types of biomass on a mill equipped with a 4 mm sieve, and despite the same grinding process conditions, each material had a different cumulative article size distribution (cPSD) [48].
The raw material degree of fragmentation has a direct and significant impact on what is produced from the granulate quality. In general, with the increase in raw material fragmentation and the following changes in PSD (decrease in the value of d 50 ), the granules' density and mechanical durability increases. This trend is confirmed both by our own research [73,74] and also by other authors [75][76][77][78][79][80][81].
It is recommended that the material for pellet production should be ground using a maximum of 4 mm mesh screen [45,82,83], whereas the recommended median particle size d 50 to obtain a high pellet strength is 0.6-0.8 mm [84], 0.5-0.7 mm [85], or even 0.2-0.4 mm [77].
The results of other studies [72,75] also indicates that not always a high degree of fragmentation guarantees higher quality granules. Bergstrom et al. noted that a material consisting of 100% of particles not exceeding 1 mm did not guarantee a granulate with a higher strength compared to a granulate made from a material where the main particle size was 4 mm and less. This is due to the bulk density (BD) of the material. Deposits made of particles with a narrow dimensional range are characterized by a lower value of BD in comparison with those with a wider dimensional range. Therefore, the criterion of proper grindability may be not only the PSD, but also a high value of BD of the obtained mixture [86]. Biomass grindability is a relatively new issue. To determine the grindability of torrefied biomass, methods used for coal are often adapted, for example, the Hardgrove Grindability Index (HGI) or Bond Work Index (BWI) [87][88][89][90]. However, even after their modification by grinding a 50 cm 3 sample instead of a 50 g sample [87], it is claimed that these indicators are not sufficiently accurate, especially when applied to untreated biomass [88]. Grindability is also described as the degree of change in PSD, d 50 , or the width of the particle size distribution called SPAN. These ratings provide for the highest possible fragmentation and unification of the particle, thus minimizing d 50 and SPAN [87,[91][92][93][94]. Such an assumption is rational from the point of view of the combustion process of pulverized fuel, but not necessary from the point of view of the compaction process. In fact, there is no good indicator of the grindability for biomass intended for the production of solid biofuels [87,88].
The grindability is influenced by both the type of material and the way the grinding process is conducted. However, there have been no studies on whether and how the drying process temperature affects grindability including are there any changes in the material, which at the grinding stage will be reflected in changes in the obtained material particle size composition or bulk density, and the effort required for the process. If the drying process allows for better grinding, it would be very important information in the development of dedicated process technologies for the production of solid biofuels from various types of biomass. Data on raw material parameters determine the selection of technological parameters of the designed granulate production lines [95][96][97][98]. The aim of this study was to investigate the effect of drying temperature on the selected biomass grindability and the energy input for the process. Biomass chosen in this study represents a wide range of raw materials used for the production of solid biofuels (i.e., woody and herbaceous biomass).

Materials
Biomass for energy purposes can have different origins and sources [1]. Due to the availability of this material, the most common energy raw material in Poland is woody biomass. However, the increasing demand for biomass has created the need to search for new sources of raw material for the production of solid biofuels. Therefore, short rotation coppices (herbaceous biomass) have become more and more popular in the energy mix. Four plant species with significantly different physical properties were used: Scots pine Pinus sylvestris L. (soft wood) and European beech Fagus sylvatica L. (hard wood) as representatives of woody biomass, and Cup plant Silphium perfoliatum L. (dicotyledonous perennial) with Giant miscanthus Miscanthus × giganteus Greef et Deu (perenial grass) as a herbaceous biomass.
All material was collected after the 2019 growth season, the woody biomass was debarked, then all samples was chipped.

Drying
A sieve shaker (LPzE-4e, Morek Multiserw, Marcyporęba, Poland) with a 400 mm diameter analytical sieve set with hole sizes of 16 and 8 mm (Morek Multiserw, Marcyporęba, Poland) was used to prepare the sample before drying, which allowed for a geometrically uniform sample. Dried was the particle size fraction that passed through the 16 mm hole size sieve and remained in the 8 mm hole size sieve (A 1 sieve class in which the particle diameter d is in the range from 8 to 16 mm). Samples of each investigated biomass were divided into three subsamples and dried at the temperatures of 60, 100, and 140 • C to the dry state. The drying process was conducted in a laboratory dryer (SLW 115, Pol-Eko, WodzisławŚląski, Poland). To determine this effect on grindability, without the influence of moisture content, further experiments were carried out on the material in a dry state.

Solid Density of Samples
After drying and cooling the samples, their solid density DE was measured using a quasi-fluid pycnometer (GeoPyc 1360, Micromeritics Instrument Corp., Norcross, GA, USA), the details of which are described in previous work [32]. The measurement was aimed at determining whether the drying temperature affects the DE value. The received results, together with the bulk density BD, were used to calculate the bed packing density ϕ.

Grinding-Preliminary Stage
The grinding process was carried out with a preliminary and main stage. First, sieve class A 1 of all materials was ground in a knife mill equipped with a sieve with a 6 mm hole diameter (Testchem LMN-100, Pszów, Poland).
For each gained material, a particle analysis was performed. The determination was carried out in accordance with EN ISO 17827-2 [99]. Each sample was sieved using a sieve set with hole size 6; 3.15; 2; 1.4 1; 0.5; 0.25 mm, and shaker (LPzE-4e, Morek Multiserw, Marcyporęba, Poland). Finally the material samples were divided into eight sieve classes (Table 1). The samples' particle size distribution (PSD) varied significantly ( Figure 1). This is mainly due to the different shapes of the ground fragments of steams (Miscanthus and Silphium) and chips (woody biomass). Before the main grinding stage, the PSD of all samples was standardized. This allowed us to avoid the influence of different input PSD on the output PSD.

of 22
Energies 2020, 13, x FOR PEER REVIEW 5 of 22 The samples' particle size distribution (PSD) varied significantly (Figure 1). This is mainly due to the different shapes of the ground fragments of steams (Miscanthus and Silphium) and chips (woody biomass). Before the main grinding stage, the PSD of all samples was standardized. This allowed us to avoid the influence of different input PSD on the output PSD. The standardized PSD included fractions B3, B4, and B5 that had the largest share in the PSD of individual samples, especially woody biomass samples. The B6, B7, and B8 fractions were omitted to prevent the large particles from causing disruption during the main grinding stage. Additionally, fractions B1 and B2 were omitted to ensure that fine particles were not only ballast during grinding. Table 2 contains the standardized PSD of all samples.

Grinding-Main Stage
All prepared samples were ground in a mill (PX-MFC 90D Polymix, Kinematika, Luzern, Switzerland) to obtain particles with a size of <2 mm. Two grinding technique was investigated: hammer and knife grinding ( Figure 2) The standardized PSD included fractions B 3 , B 4 , and B 5 that had the largest share in the PSD of individual samples, especially woody biomass samples. The B 6 , B 7 , and B 8 fractions were omitted to prevent the large particles from causing disruption during the main grinding stage. Additionally, fractions B 1 and B 2 were omitted to ensure that fine particles were not only ballast during grinding. Table 2 contains the standardized PSD of all samples.

Grinding-Main Stage
All prepared samples were ground in a mill (PX-MFC 90D Polymix, Kinematika, Luzern, Switzerland) to obtain particles with a size of <2 mm. Two grinding technique was investigated: hammer and knife grinding ( Figure 2) During the grinding process, the device was connected to the network through the ND1 energy quality measurement system (Lumel, Zielona Góra, Poland). The device measures the changes in the electric current value. Based on this, knowing the registration interval, the total energy consumption can be calculated. The volume of material ground during a single grinding cycle was 250 cm 3 . The sample to the mill was fed periodically in the amount of 25 cm 3 . The whole procedure was performed three times to eliminate the coarse measurement errors. Before milling, the idle energy demand needs to be measured. This is the energy needed to run the grinder without material in the chamber. As usual, we measured the electric current for 60 s work time of the empty grinder. We measured this parameter for both grinding systems. An average value was the reference point for net energy demand estimation, and it was also the point at which we stopped milling. During the grinding process, the device was connected to the network through the ND1 energy quality measurement system (Lumel, Zielona Góra, Poland). The device measures the changes in the electric current value. Based on this, knowing the registration interval, the total energy consumption can be calculated. The volume of material ground during a single grinding cycle was 250 cm 3 . The sample to the mill was fed periodically in the amount of 25 cm 3 . The whole procedure was performed three times to eliminate the coarse measurement errors. Before milling, the idle energy demand needs to be measured. This is the energy needed to run the grinder without material in the chamber. As usual, we measured the electric current for 60 s work time of the empty grinder. We measured this parameter for both grinding systems. An average value was the reference point for net energy demand estimation, and it was also the point at which we stopped milling.
The specific net energy demands ( . ) were calculated according to the mass of the sample that was milled.
where Enet is net energy demand and mI is sample mass. Usage of different materials requires special quantities to compare the energy demands. In the case of grinding, it is useful to know how much of the energy is used to produce the most important fraction. In our case, one very interesting parameter would be the grinding criterion first introduced by Sokolowski [92,101], which was initially used for flour production processes. This quantity is described in Equation (2) where G0.25 is the grinding criterion and S0.25 is the cumulated share of C0.25 fraction.
where S<50 is the highest cumulated share of the fraction not exceeding 50%; S>50 is the lowest cumulated share of fraction, exceeding 50%; C<50 is the sieve class corresponding to S<50; and C>50 is the sieve class corresponding to S>50. The specific net energy demands (E net.s ) were calculated according to the mass of the sample that was milled.
where E net is net energy demand and m I is sample mass. Usage of different materials requires special quantities to compare the energy demands. In the case of grinding, it is useful to know how much of the energy is used to produce the most important fraction. In our case, one very interesting parameter would be the grinding criterion first introduced by Sokolowski [92,101], which was initially used for flour production processes. This quantity is described in Equation (2) where G 0.25 is the grinding criterion and S 0.25 is the cumulated share of C 0.25 fraction. After grinding, the samples' PSD was again determined. This time, the samples were divided into seven sieve classes: C 1 : 0.1 is the particle diameter d ≤ 0.1 mm; C 2 : 0.25 is the particle diameter between 0.1 < d ≤ 0.25 mm; C 3 : 0.5-0.25 < d ≤ 0.5 mm; C 4 : 0.71-0.5 < d ≤ 0.71 mm; C 5 : 1-0.71 < d ≤ 1 mm; C 6 : 1.4-1 < d ≤ 1.4 mm; and C 7 : 2-1.4 < d ≤ 2 mm. Based on the received data, the cumulative PSD of the samples was determined. Then, the median value of particle size d 50 , which clearly shows a difference between samples fragmentation, was determined from Equation (3).
where S <50 is the highest cumulated share of the fraction not exceeding 50%; S >50 is the lowest cumulated share of fraction, exceeding 50%; C <50 is the sieve class corresponding to S <50 ; and C >50 is the sieve class corresponding to S >50 . An important indicator of the homogeneity of the mixtures is the width of the distribution SPAN and dimensionless relative SPAN [102,103]. SPAN is defined as the difference between d 90 and d 10 , where d 90 means that 90% of the mixture particle have a diameter equal or less than d 90 , respectively; and d 10 is the 10% of particles that have a diameter equal or below d 10 . Relative SPAN was calculated from Equation (4): Energies 2020, 13, 3392 7 of 22 After grinding, the BD of samples was also determined. Based on the results of measurements of DE and BD from Equation (5), the bed packing density ϕ was determined for samples before and after the main grinding stage.
where BD is the sample bulk density [kg/m 3 ] and DE is the solid density of the sample [kg/m 3 ]. In the case of the hammer system, for most materials, a slightly different course of changes was observed ( Figure 4). Accumulated curves of Pinus and Silphium showed minor changes compared to those milled with the knife system. Fagus, as for the knife system, did not show grindability dependence on drying temperature. The most significant changes for all d were observed for Miscanthus. Observed changes in cPSD indicate that an increase in drying temperature affects noticeable grindability. This is in line with the common trend observed for biomass [87,89,91]. This is probably because a drying temperature above 100 • C causes structural changes of biomass, which makes it more fragile.      Figure 5 presents a comparison of the d50 analysis of all samples. In general, the lowest values of this parameter were for Miscanthus. This indicates that this material, during milling, produces the majority of fine particles (mean value of d50 of all samples is 0.34 mm). Slightly coarser particles would be obtained for Pinus and Silphium (d50-0.39 mm) and for Fagus (d50-0.44 mm). Comparing the milling systems in most cases, the sample ground on the hammer system had a lower d50 compared to that milled on the knife system. The only exception was Miscanthus dried at 60 °C, where the trend was reversed. A similar trend was observed by Kirsten et al. by studying grass (hay) grindability [72].  Comparing the milling systems in most cases, the sample ground on the hammer system had a lower d 50 compared to that milled on the knife system. The only exception was Miscanthus dried at 60 • C, where the trend was reversed. A similar trend was observed by Kirsten et al. by studying grass (hay) grindability [72]. The greatest differences in d 50 values caused by the milling system type were observed for biomass dried at 60 • C. The increase in drying temperature resulted in a decrease between d 50 values. The exception was once again Miscanthus, this time dried at 140 • C. The influence of drying temperature on grindability is clear for Pinus, Silphium (independently of milling systems), and Miscanthus milled with a hammer system. Grindability increases (d 50 value decrease) as the drying temperature increases. In other cases, the changes were less intense. Among the tested samples, the best grindability was recorded for Miscanthus dried at 140 • C and milled on the hammer system (value d 50 of 0.3 mm). The worst was acquired for Fagus at 60 • C with the knife system (d 50 of 0.46 mm). This can be explained by the origin of the biomass. Fagus is a hard wood material with high solid density (0.66 g/cm 3 ) and Miscanthus grass has a value of 0.37 g/cm 3 . Pinus as a soft wood material has a grindability between Fagus and Miscanthus (DE = 0.51 g/cm 3 ). For these materials, the assumption that higher DE values causes a decrease in grindability is valid. In the case of Silphium, the grindability was at the Pinus level even though the DE was only 0.22 g/cm 3 . Solid density of Sylphium was lower than the value for Miscanthus, which is why the grindability of these materials is not comparable. This is probably due to the structure of Sylphium, which has a light, spongy core inside that is not very susceptible to grinding. In the cases studied, the appropriate fineness was obtained, defined by the value of d 50 , which is required for pellet production. It is recommended that it is 0.6-0.8 mm [84], 0.5-0.7 mm [85], or even 0.2-0.4 mm [77]. grindability was at the Pinus level even though the DE was only 0.22 g/cm 3 . Solid density of Sylphium was lower than the value for Miscanthus, which is why the grindability of these materials is not comparable. This is probably due to the structure of Sylphium, which has a light, spongy core inside that is not very susceptible to grinding. In the cases studied, the appropriate fineness was obtained, defined by the value of d50, which is required for pellet production. It is recommended that it is 0.6-0.8 mm [84], 0.5-0.7 mm [85], or even 0.2-0.4 mm [77]. In our case, the woody biomass did not show dependence on the milling system as well as on temperature, with an exception for Fagus dried at 100 °C and milled on a knife system. In the case of Silphium, especially in the knife system, SPAN value increased with the drying temperature, which means that the uniformity of particle size decreases. The greatest variation was found in Miscanthus samples. The material dried at 100 °C, the knife and hammer system, and the 60 °C hammer system achieved a SPAN value of about 0.46 mm, while the other variants achieved an average of 0.38 mm. This information is important from the point of view of further use of the ground material. If the material is to be used for combustion or fertilizer purposes, its dimensional uniformity is important, so a minimum SPAN value [42,87]. In the case of the pressure compaction of biomass, dimensional homogeneity is not always recommended, and it is often stressed that in a decaying material there should be particles with a wide dimensional range so that large particles would form a skeleton of the agglomerate and its fine filling [75,104,105]. Therefore, it would be necessary to determine what In our case, the woody biomass did not show dependence on the milling system as well as on temperature, with an exception for Fagus dried at 100 • C and milled on a knife system. In the case of Silphium, especially in the knife system, SPAN value increased with the drying temperature, which means that the uniformity of particle size decreases. The greatest variation was found in Miscanthus samples. The material dried at 100 • C, the knife and hammer system, and the 60 • C hammer system achieved a SPAN value of about 0.46 mm, while the other variants achieved an average of 0.38 mm. This information is important from the point of view of further use of the ground material. If the material is to be used for combustion or fertilizer purposes, its dimensional uniformity is important, so a minimum SPAN value [42,87]. In the case of the pressure compaction of biomass, dimensional homogeneity is not always recommended, and it is often stressed that in a decaying material there should be particles with a wide dimensional range so that large particles would form a skeleton of the agglomerate and its fine filling [75,104,105]. Therefore, it would be necessary to determine what SPAN value is optimal for biomass compaction processes and assess the ground material from this angle.

Grindability in Aspect of Particle Size Distribution Changes
Energies 2020, 13, x FOR PEER REVIEW 10 of 22 SPAN value is optimal for biomass compaction processes and assess the ground material from this angle.  The SPAN does not carry information about the grinding process, only about the dimensional homogeneity in the sample. To take d 50 into account, rSPAN is used. The obtained values of this parameter are shown in Figure 7. If d 50 drops and SPAN is constant, the rSPAN rises (Pinus), as d 50 takes comparable values and SPAN rises, then rSPAN also rises (Miscanthus ground with knife system at 60 and 100 • C). Therefore, due to the different nature of the change in d 50 and SPAN values, the resulting rSPAN does not give a clear indication of the change in particle size and homogeneity. Figure 6. Ground materials' SPAN value, where B is the hammer milling system and N is the knife milling system.
The SPAN does not carry information about the grinding process, only about the dimensional homogeneity in the sample. To take d50 into account, rSPAN is used. The obtained values of this parameter are shown in Figure 7. If d50 drops and SPAN is constant, the rSPAN rises (Pinus), as d50 takes comparable values and SPAN rises, then rSPAN also rises (Miscanthus ground with knife system at 60 and 100 °C). Therefore, due to the different nature of the change in d50 and SPAN values, the resulting rSPAN does not give a clear indication of the change in particle size and homogeneity. As none of the presented parameters gives clear feedback about milling quality, it is good way to provide a more complex form of data presentation. The d50, d10, and d90 should be taken into account in this assessment. Therefore, we decided to present the results obtained by combining the SPAN with d50 and the mean particle size (Figure 8) in an innovative way. The graph shows the value of SPAN (box height) and also contains information about the upper and lower dimensions of grain up to d10 and d90 from which it is calculated. Thanks to this, it is possible to compare the materials tested and also indicate which of them (despite the same SPAN value) had a larger or smaller grain size. The bars also contain information about the average grain size of 80% of the milled sample. Indication of the position of d50 in relation to the average value indicates whether 50% of grains had a smaller or comparable size to davg. Such a presentation of the results allows for unambiguous comparison of the As none of the presented parameters gives clear feedback about milling quality, it is good way to provide a more complex form of data presentation. The d 50 , d 10 , and d 90 should be taken into account in this assessment. Therefore, we decided to present the results obtained by combining the SPAN with d 50 and the mean particle size (Figure 8) in an innovative way. The graph shows the value of SPAN (box height) and also contains information about the upper and lower dimensions of grain up to d 10 and d 90 from which it is calculated. Thanks to this, it is possible to compare the materials tested and also indicate which of them (despite the same SPAN value) had a larger or smaller grain size. The bars also contain information about the average grain size of 80% of the milled sample. Indication of the position of d 50 in relation to the average value indicates whether 50% of grains had a smaller or comparable size to d avg . Such a presentation of the results allows for unambiguous comparison of the tested materials and indicates which of them is characterized by better grindability. If we know the value of d 50 of the material before milling, we can determine the relative grain reduction. In the presented study, the initial d 50 of all materials was the same, so the relative grain reduction and d 50 carries similar information and was not presented in the paper.
By analyzing the grindability plot, it can be clearly stated that the most susceptible material to grinding was Miscanthus. The obtained values of d 50 were the lowest and the particles representing 80% of the deposit were in the narrowest dimensional range (d from 0.12-0.62 mm). The material least susceptible to grinding was Fagus, the d 50 of which was the highest out of the obtained and had the widest particle size range including particles with dimensions of about 0.14-0.83 mm. The influence of the grinding system was insignificant, while the influence of temperature was most visible in the case of Pinus and Miscanthus. For Pinus, the d 50 value dropped as the temperature increased, while the SPAN changed slightly. In the case of Miscanthus, the value of d 50 decreased (hammer system) or remained unchanged (knife system), whereas the SPAN value clearly changed, reaching at least three variants (Figure 8). the widest particle size range including particles with dimensions of about 0.14-0.83 mm. The influence of the grinding system was insignificant, while the influence of temperature was most visible in the case of Pinus and Miscanthus. For Pinus, the d50 value dropped as the temperature increased, while the SPAN changed slightly. In the case of Miscanthus, the value of d50 decreased (hammer system) or remained unchanged (knife system), whereas the SPAN value clearly changed, reaching at least three variants (Figure 8).

Grindability from the Aspect of Bulk Density Changes
The changes in the bulk density (BD) is the second method used in the assessment of grindability. Figure 9 shows both BD of the material before and after grinding. It is noteworthy that although the PSD of a raw material is standardized, its BD changes, depending on the type of material and the drying temperature. Differences in BD values between materials were due to the difference in their specific densities. Comparing the changes within a given biomass type, it should be stated that in relation to the input material, the BD of the material after grinding always increased. This relation confirms that in the literature [80,106]. The highest increase was observed for Miscanthus, Fagus, and Silphium, and in the case of Pinus, the changes were insignificant. Only the Silphium Figure 8. Grindability plot where B is the hammer milling system and N is the knife milling system.

Grindability from the Aspect of Bulk Density Changes
The changes in the bulk density (BD) is the second method used in the assessment of grindability. Figure 9 shows both BD of the material before and after grinding. It is noteworthy that although the PSD of a raw material is standardized, its BD changes, depending on the type of material and the drying temperature. Differences in BD values between materials were due to the difference in their specific densities. Comparing the changes within a given biomass type, it should be stated that in relation to the input material, the BD of the material after grinding always increased. This relation confirms that in the literature [80,106]. The highest increase was observed for Miscanthus, Fagus, and Silphium, and in the case of Pinus, the changes were insignificant. Only the Silphium biomass showed a clear dependency of BD to temperature. The increase in drying temperature caused an increase in BD. A comparison of the milling systems showed that the hammer system gives a material with slightly higher BD values compared to the knife system. The exception was for Silphium, where this dependence did not occur.
In general, higher values of BD are better, especially for the agglomeration process. High BD ensures the maximization of the number of contacts between the particles in which connections are formed during agglomeration [62,107]. Therefore, the measure of grindability is the BD value of the material after grinding. In this way, we can only compare the values of materials with the same specific density. In order to make this comparison independent of DE, we used bed packing density ϕ, and the obtained results are presented in Figure 10. The specific density of the tested materials changes the drying temperature, which is most probably the effect of temperature-induced shrinkage of the material. These changes are presented in Table 3. biomass showed a clear dependency of BD to temperature. The increase in drying temperature caused an increase in BD. A comparison of the milling systems showed that the hammer system gives a material with slightly higher BD values compared to the knife system. The exception was for Silphium, where this dependence did not occur. Figure 9. Bulk density changes in the material before and after grinding; R is the initial material, B is the hammer, and N is the knife grinding system.
In general, higher values of BD are better, especially for the agglomeration process. High BD ensures the maximization of the number of contacts between the particles in which connections are formed during agglomeration [62,107]. Therefore, the measure of grindability is the BD value of the material after grinding. In this way, we can only compare the values of materials with the same specific density. In order to make this comparison independent of DE, we used bed packing density φ, and the obtained results are presented in Figure 10. The specific density of the tested materials changes the drying temperature, which is most probably the effect of temperature-induced shrinkage of the material. These changes are presented in Table 3. The results presented in Figure 10 gives more complex information than the crude bulk density. Comparing the tested materials, the highest φ values were obtained for Silphium, followed by Miscanthus, Fagus, and the lowest was for Pinus. In most cases, the value decreased with the temperature, a different course is characteristic of Silphium biomass. The influence of the grinding system was insignificant, and the biggest differences were observed for Miscanthus. Figure 9. Bulk density changes in the material before and after grinding; R is the initial material, B is the hammer, and N is the knife grinding system.
Energies 2020, 13, x FOR PEER REVIEW 13 of 22 Figure 10. Packing density φ changes in the material before and after grinding; R is the initial material, B is the hammer, and N is the knife grinding system.
The analysis of φ allows us to indicate which of the tested variants is characterized by a high gesture of packing particles in the mass and which of them is not. Bed packing density is one of the factors that can be correlated with bed structure. Silphium (temperature 100 °C, hammer system) had a value of φ 0.71. This is close to the highest regular packing of identical balls in volume (hexagonal cubic packing, hcp) in which case, φ is 0.74 [107][108][109]. On the other hand, the value obtained for random packing (rp) of identical balls was in the range of 0.57-0.68 [110]. In the case of Pinus, in the best variant, the φ only reached 0.43, which was lower than the above-mentioned values. The Fagus obtained a φ reduced to lower values of the quoted range and Miscanthus shifted to upper values. Therefore, the wood biomass showed worse deposit filling parameters and the herbaceous biomass showed better vales. This may be caused by the shape of the particles forming the deposit. In the case of Miscanthus, and especially Sylphium, the spongy core is similar to a spherical one during grinding (Figure 11). In a mixture with elongated grains (originating from grinding rest part of steams), this resulted in better filling of the available space. In the case of wood biomass, the shape of all particles is elongated. As shown by Meng et al. [110], as the elongation of the deposit-forming particles increases, the value of φ decreases To specify the changes of φ before and after grinding, it is necessary to specify the relative change of this parameter during grinding ( Table 4). The results obtained clearly indicate that the best grindability determined by the relative φ change was obtained for Miscanthus (values in the range of 28-43.4) and clearly the lowest by Pinus (4-13.3). The knife grinding system in each case caused a slight decrease in grindability compared to the hammer system. This may indicate that these grinding systems generate different particle shapes, which affects the relative φ value. An analysis based on grindability plot, which takes into account only particle size, did not show differences between the Figure 10. Packing density ϕ changes in the material before and after grinding; R is the initial material, B is the hammer, and N is the knife grinding system. The results presented in Figure 10 gives more complex information than the crude bulk density. Comparing the tested materials, the highest ϕ values were obtained for Silphium, followed by Miscanthus, Fagus, and the lowest was for Pinus. In most cases, the value decreased with the temperature, a different course is characteristic of Silphium biomass. The influence of the grinding system was insignificant, and the biggest differences were observed for Miscanthus.
The analysis of ϕ allows us to indicate which of the tested variants is characterized by a high gesture of packing particles in the mass and which of them is not. Bed packing density is one of the factors that can be correlated with bed structure. Silphium (temperature 100 • C, hammer system) had a value of ϕ 0.71. This is close to the highest regular packing of identical balls in volume (hexagonal cubic packing, hcp) in which case, ϕ is 0.74 [107][108][109]. On the other hand, the value obtained for random packing (rp) of identical balls was in the range of 0.57-0.68 [110]. In the case of Pinus, in the best variant, the ϕ only reached 0.43, which was lower than the above-mentioned values. The Fagus obtained a ϕ reduced to lower values of the quoted range and Miscanthus shifted to upper values. Therefore, the wood biomass showed worse deposit filling parameters and the herbaceous biomass showed better vales. This may be caused by the shape of the particles forming the deposit. In the case of Miscanthus, and especially Sylphium, the spongy core is similar to a spherical one during grinding (Figure 11). In a mixture with elongated grains (originating from grinding rest part of steams), this resulted in better filling of the available space. In the case of wood biomass, the shape of all particles is elongated. As shown by Meng et al. [110], as the elongation of the deposit-forming particles increases, the value of ϕ decreases To specify the changes of ϕ before and after grinding, it is necessary to specify the relative change of this parameter during grinding ( Table 4). The results obtained clearly indicate that the best grindability determined by the relative ϕ change was obtained for Miscanthus (values in the range of 28-43.4) and clearly the lowest by Pinus (4-13.3). The knife grinding system in each case caused a slight decrease in grindability compared to the hammer system. This may indicate that these grinding systems generate different particle shapes, which affects the relative ϕ value. An analysis based on grindability plot, which takes into account only particle size, did not show differences between the grinding systems. Depending on the material and grinding system, the temperature had a different effect. In the case of Fagus, the lowest values were obtained for material dried at 100 • C. Pine grindability decreased with increasing drying temperature, while for Silphium, it increased. The relative increase of ϕ, in this case, was a very good measure of grindability. It is based on the shape of the particles and changes in the arrangement in the material caused by milling. However, it does not give information about the degree of grain reduction, so the analysis of the grinding process should combine the grindability determination based on the analysis of PSD (proposed grindability plot) and ϕ changes. Correlation of these two ways of describing grindability is not possible because PSD is based on changes in grain size without taking into account their shape, while ϕ is based on the interconnection of grains of different shapes in space.  Figure 11. Shapes of the particle of the ground materials. Figure 11. Shapes of the particle of the ground materials.

Grindability from the Aspect of Energy Demands
The estimation of energy expenditure is important in assessing the profitability of technology. One of the key indicators is the net energy demand, which is the energy consumption of the process. Very often, the energy consumption is related to a unit of mass or volume. Analysis of the E net.s (Figure 12) indicates that the hammer system is more energy consuming. In any case, the increase in energy consumption compared to a knife system is at least twofold. The highest values were achieved for Miscanthus and Fagus. It is hard to determine any clear correlation of E net.s and drying temperature. In the case of the hammer system, we observed an upward trend for Fagus and a downward trend for Miscanthus. In other cases, the differences were not significant. In general, the lowest net energy demands were obtained for Silphium. For this type of biomass, E net.s was on a similar level within the milling system. At this stage, it is not possible to state the reason for these differences.

Grindability from the Aspect of Energy Demands
The estimation of energy expenditure is important in assessing the profitability of technology. One of the key indicators is the net energy demand, which is the energy consumption of the process. Very often, the energy consumption is related to a unit of mass or volume. Analysis of the Enet.s ( Figure  12) indicates that the hammer system is more energy consuming. In any case, the increase in energy consumption compared to a knife system is at least twofold. The highest values were achieved for Miscanthus and Fagus. It is hard to determine any clear correlation of Enet.s and drying temperature. In the case of the hammer system, we observed an upward trend for Fagus and a downward trend for Miscanthus. In other cases, the differences were not significant. In general, the lowest net energy demands were obtained for Silphium. For this type of biomass, Enet.s was on a similar level within the milling system. At this stage, it is not possible to state the reason for these differences. The grindability criterion is more precise in energy demands analysis. It gives information on which part of energy was used to produce the finest particles. In general, this can be treated as milling efficiency. Figure 13 presents the G0.25 in relation to the temperature and grinding system. As in the Enet.s case, better results were obtained for the knife system. There were more visible relations with temperature. In all cases, an increase in drying temperature provided positive feedback to the G0. 25 factor. This shows that materials dried in high temperature are vulnerable to producing finer particles at output during milling. Furthermore, for Silphium, there was a very noticeable relation. The grindability criterion is more precise in energy demands analysis. It gives information on which part of energy was used to produce the finest particles. In general, this can be treated as milling efficiency. Figure 13 presents the G 0.25 in relation to the temperature and grinding system. As in the E net.s case, better results were obtained for the knife system. There were more visible relations with temperature. In all cases, an increase in drying temperature provided positive feedback to the G 0.25 factor. This shows that materials dried in high temperature are vulnerable to producing finer particles at output during milling. Furthermore, for Silphium, there was a very noticeable relation. Energy demands analysis is a very complex process, in which several factors need to be considered. In this case, the energy demands to mill 1 kg of material is not an unambiguous indicator, especially when we consider the grinding quality (PSD). The G0. 25 factor gives clear information about the effectiveness of the process. A steep change in this parameter for Miscanthus (hammer system) showed that there were some changes in structure during drying in high temperatures. These  Energy demands analysis is a very complex process, in which several factors need to be considered. In this case, the energy demands to mill 1 kg of material is not an unambiguous indicator, especially when we consider the grinding quality (PSD). The G 0.25 factor gives clear information about the effectiveness of the process. A steep change in this parameter for Miscanthus (hammer system) showed that there were some changes in structure during drying in high temperatures. These changes mainly affected friability, but in the case of the knife system such tendency was not visible.
As the G 0.25 factor was established for grain milling, doubts arise about the practical mining of it in the case of biomass. It is hard to compare these two processes, as flour usually needs to be ground to a very fine powder. Biomass used in conventional pulverized boilers should be fine (below 0.25 mm) such as coal. On the other hand, there is solid biofuel production, where coarser fractions are required. This requires adaptation of the indicator to the technology needs. At this stage, it is not possible to develop such a dependency, and further research is needed.

Conclusions
The paper presents three different methods of determining biomass grindability. Standard particle size distribution analysis was extended to the form of the author's own graphical analysis. This method, in combination with the relative variability of bed packing density and grindability criterion, allowed us to make the following conclusions: • The biomass origin affects the grindability. On the basis of the grindability plot, Miscanthus was the most susceptible to grinding and Fagus was the least. Analyzing the relative changes of Miscanthus ϕ also showed the highest susceptibility, however, the Pinus biomass was the least vulnerable.
In the case of G 0.25 , the least effort should be made to grind Miscanthus with a knife system and the largest for Fagus with a hammer system (regardless of drying temperature). The exception was Miscanthus dried at 60 • C, which had the highest value of G 0.25 of all test variants • Analyzing the influence of drying temperature, the grindability plot did not indicate any noticeable trends. The relative change of ϕ, clearly correlated with drying temperature. This effect is different with the material. For Miscanthus and Silphium, the grindability increased with the temperature, in contrast to Pinus. In the case of Fagus, the best grindability was characterized by a temperature of 100 • C. Taking into account G 0.25 , in each case, we observed a decrease in the input into the process as the drying temperature increased.

•
The type of milling system, according to the grindability plot, had little influence on the course of the milling process (slightly better parameters can be obtained with a hammer system). The analysis of the relative values of ϕ changes also allowed us to state that this system will cause larger changes, and thus the milling process evaluated according to this indicator is better than with the knife system. However, the analysis of the G 0.25 clearly shows that the knife system is up to twice as energy efficient.
For each material, the drying temperature was selected individually depending on the indicator that we wanted to obtain (e.g., Silphium dried at 60 • C obtained the best results according to the grindability plot, the greatest change of ϕ occurred when the temperature exceeded 100 • C, and the lowest energy consumption was characteristic of grinding this material dried at 140 • C). According to the grindability plot and the relative ϕ change, the hammer system was slightly better, but the G 0.25 clearly indicates that the knife system was more energy efficient. The knife system was also connected with the greater susceptibility of this system to damage caused by foreign bodies in the form of stones or metal parts entering the mill.
The results in a broad and multi-aspect way describe the biomass grinding process assessed in terms of the achieved product parameters and energy consumption. Further research should focus on determining not only how the PSD, ϕ, or G 0.25 changes, but also what changes the grinding process should lead to. This means that the optimal PSD and BD of a given biomass type should be determined, depending on its use after the milling process (compaction or combustion).