4.1. Traction Decay Tests
Figure 7 shows the data from the traction decay tests for three rolling pairs with three different roughness levels: smooth (S), medium (M), and rough (R), respectively.
Figure 7a shows the evolution of the SRR during the tests. The grey background lines correspond to the raw data, and the colored lines on top were obtained by applying a moving average filter. A comparison between the initial and final roughness levels and SRRs is given in
Table 5.
Figure 7b shows the evolution of the SRR in the logarithmic scale. This plot exposes the ultra-low initial SRR measured with the rough rollers and its increase with time.
Figure 7c shows the contact force
and traction force
, which were maintained at a stable level for all roughness levels. Finally,
Figure 7d shows the applied braking torque
, the tractive torque
, and the frictional torque
. Note that the braking torque
is shown as negative only for the sake of clarity since its value is very close to that of the tractive torque
. This means that the contribution of the needle bearings to the total resisting torque is minimal (Equation (
1)). In fact, the average frictional torque is
and accounts for only ≈2.6% of the total resisting torque. The average frictional torque
generated by the needle bearings corresponds to an average friction coefficient
.
During the traction decay tests, for all roughness levels (i.e., S, M, and R), a substantial increase in slipperiness was recorded. The rough and smooth surfaces exhibited the smallest and largest SRRs, respectively. However, in
Table 5, it is interesting to see that, proportionally, the increase in the SRR was practically the same for both roughness levels. The most significant rise in the SRR occurred with the medium roughness level, where the SRR increased 17.9 times during the test. This considerable increase in slipperiness can be attributed to the loss of roughness caused by polishing wear. The change in roughness for the medium roughness level was the largest compared to the other two (
Table 5). Here, it should be noted that the initial SRR (i.e., 0.57%) in
Table 5 does not match that in
Figure 7b. The reason for this is that 0.57% corresponds to the initial SRR recorded before applying a moving average filter. The same explanation applies to the smooth and rough rollers.
For the rough rollers, a (smaller) decrease in roughness was also observed. However, for the smooth surface, the roughness “increased” during the test. This was attributed to the occurrence of severe adhesive wear, which also explains the large fluctuations in the SRR in
Figure 7a. The sudden decline in the SRR in minute 75 and the subsequent fluctuations were linked to the occurrence of severe adhesive wear.
Figure 8,
Figure 9 and
Figure 10 show the initial and final surface topologies of the rollers with rough, smooth, and medium roughness levels.
It should be pointed out that while we recognize the importance of achieving repeatable results, our primary goal in this study was to present our methodology and validate our novel tribometer. Consequently, we stopped ourselves from diving into the details of adhesive wear initiation and the repeatability of such outcomes since conducting such an investigation would demand a substantial number of rolling pairs and a different research design. However, based on the patterns observed and the insights gathered during testing, we believe that comparable outcomes may be anticipated when repeating the experiment with a smooth surface. This anticipation holds valid particularly for smooth stainless steel rollers, given the susceptibility of the material to suffer from adhesive wear.
4.2. Torque-Mode Traction Tests
Figure 11 shows the data acquired to construct one traction curve. In this case, curve (5) according to the sequence in
Table 4. It should be noted that in
Figure 11, only two legends are shown due to limited space. They correspond to the first and last steps followed, from low to high braking torques
, to generate different SRRs. The color of the lines can be used as a guide to distinguish torques and SRRs from different steps.
In this section, the procedure to construct a traction curve is further described, with aid from the results obtained.
Figure 11a shows the steps taken in the braking torque
from low to high. At each step, the braking torque
was kept stable for 60
, while speed, force, and torque signals were acquired.
Figure 11b shows the SRRs recorded during the test for each applied braking torque
. Low SRRs correspond to small braking torques and vice versa. For example, for the largest average braking torque
, the mean SRR is 17.8%. It is worthwhile pointing out that fluctuations in the SRR were recorded at larger braking torques in particular. The amplitude of these fluctuations increases proportionally with the braking torque
. At low SRRs, the fluctuations are too small to be noticed. This intriguing behavior can be better explained by examining curve (5) in
Figure 12b (at
. At small resisting torques,
values, the slope of the curve is close to zero. Therefore, small fluctuations in the resisting torque
cause much smaller fluctuations in the SRR. However, at higher resisting torques
, the slope becomes much steeper. Under these conditions, small fluctuations in the resisting torque cause noticeable variations in the SRR. At
, the slope of the curve is close to 1. This means that small torque fluctuations cause very large changes in the SRR. Under these conditions, the contact is highly unstable, and full sliding (i.e., SRR = 200%) can suddenly occur. This argument also explains why torque control is attractive, particularly for conducting tests at small SRRs, close to the linear region of the traction curve.
The increasing amplitude of the SRR fluctuations at higher resisting torques is an attention-grabbing aspect. An escalating amplitude could be used to indicate that the limiting traction that the interface can handle is about to be reached. Furthermore, the frequency of the SRR fluctuations (
Figure 11b) also appears to increase at higher torques, where more slippage occurs. This phenomenon might be linked to viscosity changes due to thermal effects as follows: At low SRRs, the contact cools down, viscosity increases, and asperity contact is lost; thus, the SRR goes up. At higher SRRs, the contact temperature increases, and viscosity drops, allowing for more contact between asperities, and hence, the SRR goes down. This behavior seems to follow a repeating and predictable pattern, making it an intriguing topic for further investigation.
In addition, the following should be highlighted. Although the braking torque was increased in steps as small as for constructing the traction curves, much smaller increments can be made if required. As a result, the number of average data points can be increased, yielding traction curves with improved resolution. In fact, the CRT allows us to apply braking torques with increments as small as . Concerning the stability of the braking torque, under closed-loop regulation, variations of roughly ±0.05% of the set torque value can be anticipated.
Figure 12 shows nine torque-mode traction curves for three different contact forces (
,
,
) and three different rotational speeds (50, 150, and 450 rpm). On the right side, the logarithmic plots depict the ultra-low SRRs generated at small resisting torques. Using curve
Figure 12b (at
and
) as an example, it can be seen that it contains data points at ultra-low SRRs ranging from 0.015% to 0.11% and a total of 15 data points from 0 to 1%. This demonstrates the advantage of operating in torque mode when generating high-resolution traction curves that include ultra-small SRRs.
Overall, it can be concluded that the SRR increases as a function of the resisting torque
, except in cases where traction is “enhanced” due to adhesive wear damage. At higher contact forces, the interface can tolerate higher resisting torques at small SRRs. At higher speeds, the tractive properties of the interface decrease due to a reduction in the contact between asperities produced by a thicker lubricating film. For the plots on the left side, the vertical lines 1, 2, and 3 can be used to highlight the influence of speed. For example, in
Figure 12c, line 2 intersects the
curve at an SRR of 9%, in
Figure 12b, at 6%, and in
Figure 12a, at 3%. This indicates that at higher speeds, more slippage is necessary to balance the same resisting torques.
It is worth emphasizing that the limited kinetic power rating of a magnetic hysteresis brake can become a constraint. This rating is determined by both the operating speed and braking torque. For the HB1750 model, under continuous operation it should not exceed 350 . However, for a brief 5 period, the maximum allowable kinetic power is 1200 . During the tests conducted at 450 rpm, in the last four steps with braking torques above Nm, the kinetic power exceeded 350 . In the final and most critical step, the kinetic power reached 400 . Notably, these last steps resulted in a significant temperature increase in the brake. Under these circumstances, a fan or compressed air can be used on the side of the brake to improve the dissipation. However, it is essential to be cautious to prevent overheating. In addition, these constraints should be considered when establishing the parameters for testing.
To conclude this section, it is important to point out that the data from the curves in
Figure 12 can also be plotted in a conventional way, with the SRR in the
x-axis and the traction coefficient
(or the tractive torque
) in the
y-axis. In fact, if the curves in
Figure 12 are mirrored on the right and then rotated
clockwise, the plots become conventional traction curves, as shown in
Figure 13.