Coefficient of Friction
Experiments, with a tribology cell connected to a rheometer, have successfully been conducted in several research areas. Goh and coworkers  have analyzed the food texture and the lubrication properties of emulsion systems using a tribology cell and they conclude that it can function as an appropriate investigating tool. Furthermore, dairy products in the form of semisolid or liquid , as well as various greases  can likewise be analyzed with success using the tribo-rheometer system. However, tribological measurements with a rheometer on coated abrasives have until now not been tested. The results from the measured COF experiments performed with the rheo-tribometer, are depicted in Fig. 3.
The initial experiments were performed to investigate the validity of the measurement technique. Because tribological systems can be very complex, it is demanding to obtain credible results from such experiments. Steel balls were grinded on MI231A coated abrasive material and the COF was continuously measured and plotted versus the sliding distance. Seven separate tests with coated abrasive discs and new steel balls were performed. The friction coefficient was calculated and recorded every two seconds using the Anton Paar RheoCompass software. At the start of the test all friction curves presented similar friction behavior and the COF increased up to a sliding distance of about 70 m (Fig. 3). This behavior is typical in similar friction measurements and has been attributed to a “Run-in” phase . During this phase the highest contact areas break loose and the surface topography changes. After the Run-in phase, the COF generally stabilized into a “Steady state” phase. In these experiments, the steady state lasted between a sliding distance of roughly 70 and 220 m, and afterwards transcended into a so-called “Failure” phase. The Failure phase was recognized by increasing COF, as well as more significant variations in the COF of the different samples after a sliding distance of 220 m, as shown in Fig. 3. The presence of wear debris in the contact area between the coated abrasive and the steel balls will most likely make the tribological characteristics unstable, which has also previously been demonstrated . An average COF was calculated from the different measurements, which is shown as the bold line in the graph. The average standard deviations were 1.0% during the Run-in phase, 1.5% during the Steady state phase, and 3.5% during the Failure phase.
Results from previously published studies show that wear tests can vary greatly depending on different parameters in the tribological system [32, 33]. The COF factor of any wear test method is generally used to rank the grinding material composition, and therefore it is very important to validate the repeatability of the wear performance, determined through different test setups. According to  the interpretation of the COF values can be challenging. The COF is very dependent of system parameters, therefore the interpretation of the results should always be done with caution. With these initial COF measurements, it was observed that the grinding procedure consisted of three distinct phases: Run-in, Steady state and Failure. The calculated standard deviations of the three different phases also show the increased uncertainty of COF measurements performed during the Failure phase. After the Failure phase, the friction coefficient might reach another steady state plateau before the coated abrasive is completely worn out according to literature . Consequently, due to the high uncertainties within the Failure phase, the focus should not be on this phase, but rather on the Steady state phase when comparing the COF factor of different abrasive materials.
After the experiments shown in Fig. 3, the morphology of the worn abrasive (MI231A) was examined using a stereomicroscope and a scanning electron microscope. The objective was to investigate more closely the wear of the coated abrasive and to examine in detail why these three phases, and especially the Failure phase, occur during grinding. In this process, there are two phenomena that are of interest. Firstly, the appearance of wear debris in the system, and secondly the wear of the coated abrasive. An unused coated abrasive surface can be seen in Fig. 4a and is compared to Fig. 4b which depicts the coated abrasives after the experiment. In Fig. 4b, the abrasive particles and the sandpaper are completely worn out. Figure 4c shows the wear path made by the tribology cell, which clearly is full of debris. These two phenomena explain why the coefficient of friction enters a Failure phase in the final step of the wear process. However, it must be pointed out that this grinding behavior with steel balls against abrasive material causes major forces upon the abrasive material, which does not happen in real-life applications.
Effect of Normal Load on the Wear Mechanism
To investigate whether different loads of pressure have an impact on the wear process in our experimental system, loads between 1 and 10 N were applied by the rheometer upon the abrasive material MI231A P800 (20 μm). Figure 5 shows how different loads affected the mass loss of the steel balls and the Gap. The mass loss was calculated by weighing the balls before and after the wear test. The results clearly showed that the mass loss increased with increasing normal load, which indicated an enhanced wear. When analyzing the behavior of tribo-rheometer systems, the Gap is often noted as an important parameter in the analysis . The Gap can be obtained directly from the RheoCompass computer software and could presumably be an indicator of abrasion. During the test when the tribology cell head is in contact with the interface, the head moves up and down to regulate the normal force to the set value. The Gap depends on the surface as well. On a hard (non-deflecting) surface, the Gap is roughly equivalent to wear depth, while on a soft surface, the Gap includes both wear and deflection.
The corresponding change in the Gap serves as an indicator of the transition and consequently, the wear performance. As more pressure is applied in the system, the Gap is enhanced, which corresponds well with its proposed role as an indicator of wear (Fig. 5). These results suggested that this test methodology is appropriate for testing the tribological characteristics of MI231A wear on steel.
The COF factor was examined at steady state, since this phase seemed to be the most reliable for comparing different coated abrasives (Fig. 3). The results show a slight increase in COF with increasing normal load.
Different factors are believed to contribute to the disparity of the wear mechanism. These include the shape, the size, and the chemical composition of the grains, but also the applied load in the grinding system is of great importance. Misra and Finnie showed that the wear of a specimen increased with increasing normal load during the abrasion process . Furthermore, in a study examining the wear mechanism in two-body abrasion using a pin-on-disc test rig on silicon carbide paper, the wear rate was examined using three different sizes of grains; 5 μm, 15.2 μm and 82 μm . The results showed that the wear rate on steel-pins is increased with all the grain sizes at a normal load from about 2 N up to 25 N. In addition, normal load studies on other materials, for example on various aluminum alloys  or experiments performed at higher applied normal loads (10–40 N) [14, 33], show that the wear increases with increasing normal load during the abrasive process.
Effect of Grain Size
Wear tests on different abrasives with various grit sizes were performed to determine whether the rheo-tribometer could be utilized to study this well-known size effect phenomenon. The results in Fig. 3 suggest that the Failure phase might not be the most optimal phase for analyzing the COF factor when comparing different coated abrasives with each other. However, to get a measurable mass loss from the steel balls in the rheo-tribometer, the tests proceeded to 500 m.
It has previously been demonstrated that the size of the abrasive grains has an effect on the abrasion wear mechanism [20, 35,36,37]. Consequently, for small particle sizes an increase in the wear rate is observed as the particle size increases. However, after a critical grain size at about 100 μm, there is no further significant change in the wear rate . Figure 6a shows experiments performed on coated abrasives with grain sizes between 9 and 40 μm.
To determine if the wear process in our tests could distinguish the known size effect, the abrasive mass loss was calculated by weighing the steel balls before and after the wear test. The mass loss of the steel balls, grinded by the different abrasives, is shown as grey bars in Fig. 6a. The results showed that the wear on steel is enhanced with increasing grain sizes, which was in agreement with previously reported findings [20, 36, 37]. The solid line in Fig. 6a, depicts the change in the Gap, measured by the RheoCompass software. The Gap increased with increasing grain size, which corresponds well with the proposed size effect when using small grain sizes . The COF factor measured at steady state also increases with increasing particle diameter. Hence, based on these results and the results in Fig. 5, it is concluded that the Gap can be used as a reliable indicator of the wear performance.
A separate validation experiment was performed to verify the behavior of the tested abrasive materials. To do this, a special constructed test robot, which is normally used at Mirka for internal wear tests, was utilized. According to an internal standard test method, the abrasion process of the abrasives on steel was analyzed. Before and after the wear tests the specimens were weighed and the mass loss was calculated. Figure 6b depicts the sample loss as a function of grain size. Although the wear process is more effective in this test (more sample loss) in comparison to the tribology cell test method, the trend showed similar behavior, i.e. the wear increased with increasing grain size. These results further confirmed that the method for testing abrasives utilized in this study is working as proposed.
The accuracy of a test method is crucial. The standard deviation usually gives a good indication of the stability of the experimental setup. The results presented in Figs. 5 and 6 are based on three individual tests, and consequently show different accuracies. As can be seen in Fig. 5, the standard deviations between the triplicate samples within the gap experiment exhibit relatively small standard deviation values. One possible reason for this could be that the Gap was directly obtained from the software, and extra steps were therefore avoided. The standard deviations in the experiment with the mass loss of steel balls (Figs. 5 and 6) are relatively large between the triplicate samples. This might be due to the complexity of gravimetric analysis. Although weighing samples is an appropriate method in many cases, the number of steps and other insecurities in such experimental procedures can be of importance. The magnitude of the standard deviations could also be due to the placement of the abrasive specimens in the jumbo reel, which is a phenomenon that has also previously been observed .