Abstract
Groundwood pulping is a process in which logs are pressed against a rotating grinding stone. A conventional grinding stone is generally made of grinding particles in a vitrified matrix. As the particles are practically round, their contact with the wood is limited to occasional point contacts. The interaction between the particles and the wood occurs at random positions and at random times, only intermittently contributing to the defibration process. In this work, well-defined grinding tools with asperities giving line contacts rather than point contacts were tested. The tool surface asperities were elongated in shape and positioned with different density over the surface. The tools were tested in a lab-scale equipment at elevated temperatures, and their performance was compared to that of a conventional grinding stone. The grinding mechanisms varied between the different tools, and the specific grinding energy was reduced compared to the conventional tool.
Download PDF
Full Article
Evaluation of Well-defined Tool Surface Designs for Groundwood Pulping
Magnus Heldin * and Urban Wiklund
Groundwood pulping is a process in which logs are pressed against a rotating grinding stone. A conventional grinding stone is generally made of grinding particles in a vitrified matrix. As the particles are practically round, their contact with the wood is limited to occasional point contacts. The interaction between the particles and the wood occurs at random positions and at random times, only intermittently contributing to the defibration process. In this work, well-defined grinding tools with asperities giving line contacts rather than point contacts were tested. The tool surface asperities were elongated in shape and positioned with different density over the surface. The tools were tested in a lab-scale equipment at elevated temperatures, and their performance was compared to that of a conventional grinding stone. The grinding mechanisms varied between the different tools, and the specific grinding energy was reduced compared to the conventional tool.
Keywords: Groundwood pulping; Diamond tools; Energy consumption; Tomography; Grinding mechanisms
Contact information: Ångström Tribomaterials Group, Division of Applied Materials Science, Department of Engineering Sciences, Uppsala University, Box 534, 75121 Uppsala, Sweden;
* Corresponding author: magnus.heldin@angstrom.uu.se
INTRODUCTION
Groundwood pulping is a mechanical pulping process in which wood logs are pressed against a rotating grinding stone to separate the fibers, creating pulp. The energy consumption is high in these processes, and most of the energy is lost through undesired viscoelastic deformation of the wood; only a minor part of the energy is beneficial to the separation of the fibers (Campbell 1934; Uhmeier and Salmén 1996). During grinding, the groundwood pulping processes produce pulp of all different fractions, from long slender fibers to fines. Controlling the characteristics of the pulp is desirable when making pulp to be used in different applications, as the requirements will differ. For instance, the fibers preferred when producing characteristics used in newsprint paper are not that suitable for reinforcement in polymers.
The process parameters have a strong influence on both the process itself and the properties of the produced pulp. The temperature at which the grinding takes place is probably the most well-known and most studied parameter. Softening of the wood will occur as the temperature is increased and, in the range of temperatures used in pulping processes, the softening hemicelluloses and lignin will affect the mechanical properties of the wood (Koran 1981; Blechschmidt et al. 1986) and facilitate defibration.
Other influential parameters are the properties of the tool. Conventional grinding stones consist of grinding particles fixed in a matrix, typically alumina particles in a vitrified matrix. The particles are randomly distributed over the surface of these grinding stones, making the occasional contacts between grinding particles and wood fibers very stochastic in nature.
The size and shape of the particles are important in the defibration process, affecting the mechanisms in the contact between the grinding stone and the wood, and changing the energy consumption of the process as well as the quality of the fibers produced (Enström et al. 1990; Sandås and Lönnberg 1990; Sandås 1991a,b; Lönnberg et al. 1996; Tuovinen et al. 2008). Fatigue is believed to be an important part of the defibration (Salmi et al. 2012a,b) and by profiling the grinding stone, i.e., creating a sinusoidal topography along the circumference of the stone, fatigue can be enhanced in the wood during grinding, while also reducing the energy consumption (Björkqvist and Lucander 2001; Björkqvist et al. 2007). The surface pattern can be used to influence the character of the released fibers. Tools with a serrated profile across the width of a conventional grinding stone have been shown to increase the fraction of fines in the pulp (Nurminen et al. 2018).
The distribution of particles in the grinding stone surface can be changed from random, as in conventional stones, to controlled (Tuovinen et al. 2009; Tuovinen and Fardim 2015) allowing greater control of the particle interactions with the wood during grinding.
Most previous research on wood grinding has employed grinding tools where, in some form, particles are fixed in a supportive matrix. This has generally meant that the particles are round in shape and their interactions with the wood are in the form of point contacts. However, extremely well-defined grinding surfaces have been produced using diamond deposition on patterned silicon and have been used to grind other types of materials (Gåhlin et al. 1999). The same production technique has been used to create embossing tools with surfaces upon which the asperities had an elongated profile (Pettersson and Jacobson 2006). In this work, grinding surfaces with surface asperities designed to give line contacts, instead of point contacts, were produced. The use of such tool surfaces has the potential of not only affecting the energy consumption in pulp production, but also the fibrillation and transport of fibers out of the grinding zone. This means that the tools can be tailored to produce fibers with different characteristics.
This work investigated the influence of three different well-defined grinding surfaces on mechanisms and energy consumption during grinding at different temperatures in a lab-scale grinding setup. A piece of a conventional grinding stone was used for comparison.
EXPERIMENTAL
The equipment and method used in this work are described in greater detail in a previous work (Heldin and Wiklund 2019) and will only briefly be described here.
Grinding Experiments
A setup similar to a lathe was used for the grinding experiments. Flat patterned grinding tools were pressed against a rotating cylindrical wood specimen by a spring-loaded tool holder, instrumented to measure the normal and tangential forces affecting the tool. The setup was placed inside a pressure chamber to allow for a test environment similar to the industrial grinding process, which is carried out at elevated pressure using steam. The test parameters used are displayed in Table 1. After each test the released fibers were collected for analysis.
The well-defined grinding tools used were all 5 × 5 mm2 and made of nanocrystalline diamond films soldered to steel backings. One tool surface consisted of long parallel ridges (SLong) that were 89 µm in base width, 64 µm high, and spanning the full width of the tool, as shown in Figs. 1a and b. A second tool surface with a sparse pattern of shorter ridges (SSparse) had 500 µm long, 86 µm wide, and 61 µm high ridges distributed over its surface (see Figs. 1c and d). The third tool surface, with a dense pattern of short ridges (SDense), had 260 µm long, 90 µm wide, and 65 µm high ridges placed rather close to each other, see Figs. 1e and f. The patterns present differences in both asperity density and contact lengths as well as different possibilities to transport fibers out of the contact.
For comparison, a piece of a conventional grinding stone was obtained from a Swedish pulp mill and cut into a 5 × 5 mm2 piece, see Figs. 1g and h.
All wood specimens used came from a 50-year-old Norway spruce grown in the middle of Sweden. The wood was preserved by storing it in a freezer directly after cutting. Prior to testing, the wood specimen was placed in the test chamber and heated in a steam atmosphere to reach the desired temperature, taking about an hour. Different tests were performed in sequence along the length of a specimen, avoiding knots and other defects in the wood. After completing the last test on a wood specimen, it was left to dry in a lab environment, i.e., 20 °C and 50% RH.
Table 1. Test Parameters Used in the Grinding Tests
Analysis
The removed amount of fibers was too small to rely on fiber collection and weight measurements for its quantification. Instead, an image-based analysis of the cross section of the grinding tracks was employed to quantify the amount of fibers removed and allow for comparisons between the tests.
When the wood had dried, a 5 mm long section of each grinding track was cut at the position of the circumference where the annual rings were perpendicular to the specimen surface. All analysis was then performed on these sections.
Computed tomography (µCT) was used to measure the area of removed fibers from the cross section of the tracks. It was performed using a Skyscan 1172 (Bruker microCT, Kontich, Belgium) run at 53 kV source voltage and 188 µA source current. The acquired radiographs had a resolution of 1.97 × 1.97 µm2/pixel. From these, the removed area was measured in 50 reconstructed cross sections, spanning a total of 2 mm along each grinding track.
The produced fiber volumes were small, making measurements of conventional pulp properties, such as freeness, impossible. Instead, micrographs of the grinding track’s surface and of the released fibers were captured using a Zeiss Leo 1550 scanning electron microscope (SEM; Jena, Germany) to investigate the mechanisms of fiber separation and the characteristics of the separated fibers. Prior to the SEM imaging, wood and fiber samples were coated with a thin gold/palladium film to reduce surface charging in the microscope.
Fig. 1. Parts of the well-defined grinding surfaces; (a-b) long ridges across the whole tool, (c-d) ridges sparsely placed, and (e-f) densely placed. Panels (a), (c), and (e) show the tool from a top view and (c), (d), and (f) show the corresponding tools in cross section. The tool asperities all have a triangular cross section and an apex angle of about 70°. A small piece of a conventional grinding stone with grits about 200 µm in diameter, depicted in (g) top view and (h) in cross section
Grinding Tool Surface Profiles
The four tools present different surfaces to the wood. To be able to relate the grinding tool performance to the surface structure, load bearing curves were produced for all four tools. For the diamond tools, the geometries of the ridges were measured from SEM images, and from these, the load bearing curves were calculated.
The grinding stone surface was analyzed using a ZYGO Nexview NX2 Optical profiler (Middlefield, CT, USA), stitching several images obtained using a 10x magnification lens. Due to poor reflectivity in some areas, the final analysis was made on a 4 × 4 mm2 area of the grinding stone surface, and a load bearing curve was calculated using the ZYGO Mx software.
Grinding Energy Calculations
The specific grinding energy during each test were calculated from the tangential force and removed fiber cross-section areas, with details presented in a previous work (Heldin and Wiklund 2019). Equation 1 was used to calculate the mass specific grinding energy, Em,
Em = FTωrt/2πrAρ = FTωt/2πAρ (1)
where FT, is the tangential force, ω is the angular velocity, r the radius of the workpiece, t the test time, A the cross-section area of removed fibers, and ρ the density of the wood.
As the angular velocity in the test is 2π rad/s and the test time is 120 s, one can simplify the equation as follows:
Em = 120FT/Aρ (2)
Assuming a wood density of 400 kg/m3, the specific grinding energy was calculated using the average measured tangential force and average track cross-section area in each test.
RESULTS AND DISCUSSION
The tangential force showed a changing behavior in the beginning of most tests, lasting some tens of seconds, especially at the lowest temperature (Fig. 2), similar to a running-in behavior of sliding mating surfaces. After this, the tangential force generally stabilized.
Fig. 2. Representative examples of tangential forces from tests run at (a) 70 °C, (b) 90 °C, and (c) 110 °C
The average tangential force varied between tests for any specific tool and temperature (Fig. 3), but on average all four grinding tools showed reduced average tangential force with increased temperature, especially when going from 90 to 110 °C. The degree of reduction was different for each tool, with SLong showing the largest reduction in tangential force, and SDense and the grinding stone showing only a slight reduction.
Fig. 3. Average tangential forces for all 60 tests performed
The µCT cross section images in Figs. 4 and 5 show that SLong, SDense, and the grinding stone show similar trends with increasing areas of fiber removal as the temperature increased.
Fig. 4. Examples cross section images obtained using µCT. The grinding tools have moved perpendicular to the image plane and the areas of removed fibers have been highlighted. All images have the same magnification and the tracks are all 5 mm wide.
Fig. 5. Measured area of removed fibers from the µCT cross section images, with a total of 50 measurements for each sample
SSparse showed a large increase in area as the temperature increased from 70 to 90 °C, but then not much larger when increasing the temperature to 110 °C.
The calculated specific grinding energies for the tests (Fig. 6) showed some scatter between the individual tests performed with a specific tool. The spread was large at 70 °C, but much smaller at higher temperatures. For every tool, however, the trend was similar in that the specific grinding energy was reduced as the temperature increased. The lowest specific grinding energy was measured during grinding with SLong at 110 °C, followed by SSparse at 110°C.
Fig. 6. The specific grinding energy for each of the 60 tests (a) and the median specific grinding energy for the different surfaces at three different temperatures (b)
SEM micrographs of the grinding tracks (Fig. 7) look decievingly similar at first sight, but there are differences. Generally, for all grinding tools the length of the partly separated fibers, i.e., of those still attached to the wood, increased with increasing temperature. SLong and the grinding stone both showed long partly separated fibers at all temperatures. At the lower temperatures, however, the grinding stone tended to cut the fibers, visible in the micrograph from the 70 °C tests in Fig. 7. Additionally, SSparse and SDense left grinding tracks with many similarities. When grinding at 70 and 90 °C, both surfaces generally had a low amount of partly separated fibers, and those present were short and fractured.
Images of the released fibers, shown in Fig. 8, reveal that at the lower temperatures all tools produced short fibers together with fragments and fines. The fines tended to gather into bundles or rolls, especially the fines produced at 70 °C. For all four tools, the length of the fibers increased with increasing temperature. For the diamond surfaces, a change in fiber character was found only at 110 °C, whereas for the grinding stone there was a gradual change when going from 70, to 90, and 110 °C. At 110 °C all tools produced long fibers with a lower amount of small fragments and fines. The least damaged fibers were obtained with SSparse.
The load bearing curves, see Fig. 9a, show that the grinding stone has a much deeper surface profile than any of the diamond surfaces. In Fig. 9b, rescaled to visualize the differences between the diamond surfaces, the surface profiles of SLong and SDense are very similar and clearly different from SSparse. These differences are further illustrated in Fig. 10, showing that at identical load bearing areas, here shown as 15%, the shape and distribution of the areas carrying the load are very different for the respective tools.