3.1. Hysteresis Loops at Different Reduced Frequencies
In this section, hysteresis loops of the lift coefficient and their difference at three reduced frequencies, listed in
Table 2, in clean air and air–particle flows are discussed. All cases are computed on a 12-core workstation with a CPU frequency of 3.4 GHz.
Figure 4 shows the hysteresis loops of two airfoils, in four cases, in a clean air flow. For NACA0012 airfoil, the lift coefficient has a linear change with the angle of attack in the steady case, and the values during the up stroke are approximate to those during the down stroke in every pitching case. Three loops also have little difference, except for the inclined angle. However, for S809 airfoil, according to the analysis of the steady case by Guo et al. [
34], the lift coefficient stops linearly increasing and decreases slightly near the angle of attack of 8°, due to the forward movement of the boundary-layer separation point from the trailing edge, which is called the light stall. During the following stroke, the lift coefficient increases again and drops into deep stall at an angle greater than 16°. In three pitching cases, the reduced frequency has large effect on the hysteresis loop. The lift coefficients during the up stroke are greatly different from those during the down stroke, which will be discussed as follows, and the mechanisms are also revealed.
In the low reduced frequency case (Case 1), the loop during the up stroke has similar trend with the steady results. As shown in the
Figure 4b, the loops are divided into several segments for comparison purposes. Due to the hysteresis effect, a larger range of the linear segment is obtained, and the segment ab has a smaller range than that of the steady case. In segment bb
1, the lift coefficient increases again because of the higher suction peak, which is shown in
Figure 5. With the angle of attack increasing, the leading-edge vortex and suction peak barely change, the boundary-layer separation point goes forward rapidly. At the same time, the continuously decreasing pitching speed also cause the lift coefficient to approach the value of the steady case. They both lead to the decrease of the lift coefficient in segment b
1b
2. Then, a similar change to the steady case is presented, until the angle of attack reaches to its maximum value. In the segment de, the lift coefficient keeps close to the steady result because of the slow pitching speed. Notably, the light stall is still effective in this segment. With the further decrease of the angle of attack in segment ef, the airfoil no longer suffers the effect of light stall, the boundary-layer reattaches, and the lift coefficient increases. Finally, the linear change reappears at angles of attack smaller than 8.1°, and the lift coefficients are larger than those during the up stroke.
In the moderate reduced frequency case (Case 2), the linear segment ends up at a larger angle of attack and the segment ab disappears, due to the greater pitching speed. Under the effect of light stall, the lift coefficient in segment ac increases at a smaller speed. Then, in the segment cd, the leading-edge vortex sheds, and the lift coefficient drops sharply, which is defined as deep stall (or full stall). In the following segment, the lift coefficient slowly decreases to the initial value of the up stroke.
In the high reduced frequency case (Case 3), the range of linear segment further increases and the segment ac becomes smaller. There are only two segments during the whole up stroke, and the deep stall also disappears, due to the high pitching speed. That is because the virtual camber effect, which was discussed by Hu et al. [
16], starts to dominate the flow field. In addition, the fluctuations in case 2 and 3 are caused by the small number of the inner iteration per step, which is described in
Figure 2a. In the following discussion, the fluctuation will become larger, due to the hysteresis effect, especially in the loop of the relative increment between clean air and air–particle cases. Therefore, the mean value is more accurate.
Next, the cases in the air–particle flow are added to the comparison. For the three cases above, increment loops of the lift coefficient of two airfoils, caused by particles, are presented.
Figure 6 and
Figure 7 display the loops for Case 1 at different particle mass concentrations and the relative increments between concentrations of 0% and 1%, respectively. For NACA0012 airfoil, particles have little effect on the shape of the lift coefficient loop but a decrease for its inclined angle. The flat loop of the relative increment indicates the small loss in aerodynamic performance throughout and its little difference between up and down strokes. The peak value of −2.016% is obtained at the maximum angle of attack. For S809 airfoil, the relative increment loop is also divided into several segments, as well as the discussion in
Figure 4b. During the up stroke, particles make the linear segment end up at a larger angle of attack, which increases with the particle concentration. The segment ab is obtained at larger angles of attack; its relative increment decreases first and then increases. In these two segments, the relative increments are both small enough to be out of consideration. Due to the large difference between increments at Point b and Point b
1 by particles, the change of segment bb
1 strongly depends on the particle concentration. It means that the angle of attack, at which the maximum lift coefficient is obtained, has two possibilities. As shown in
Figure 6b, the lift coefficient of segment bb
1 increases with the angle of attack at the particle concentration of 0%, and a decrease is presented at the concentration of 2%. The concentration of 1% is the critical value. Then, the maximum lift coefficients at concentrations of 0% and 2% are 1.127 and 1.087, which are obtained at angles of attack of 10.5° (Point b
1) and 9.25° (Point b), respectively. In this segment, the relative increment rapidly increases. In addition, particles make the Point b
1 move to a smaller angle of attack and the larger the concentration, the larger the movement distance and smaller the range of segment bb
1. In the following three segments, the lift coefficient in the air–particle flow is approximately parallel to that in the clean air flow. The relative increment continues to increase, and the maximum value of −6.55% is obtained at the angle of attack of 11.3° in segment b
1b
2. In whole segment b
2c and the first half of segment de, the relative increment is around 3%, and when the angle of attack decreases to 10°, it increases to a peak value of −4.5%. Then, the increment rapidly decreases, until the positive value is obtained. Compared to that in the clean air flow, the Point E appears at a larger angle of attack in the air–particle flow. It means that particles can advance the reattach of the boundary layer. The range of segment ef increases with the particle concentration because the influence of particles at Point e is larger than that at Point f. Due to the hysteresis effect, two loops (m = 0% and 1%) intersect in this segment and a positive peak relative increment of 2.16% is obtained at the angle of 9.06°. Then, it decreases until the negative value is obtained again. Finally, in the linear segment, the relative increment changes in a negligible range, although it is larger than that during the up stroke.
In
Figure 8 and
Figure 9, for Case 2, particles also have little effect on the loop of NACA0012 airfoil. The angle of attack, where the maximum lift coefficient is obtained, is unchanged, which is different from that of S809 airfoil. During the up stroke, the relative increments of two airfoils are around −0.25%, except in the last section. From 12.5° to 13.5°, they increase to −1.3% and −6.85%, respectively. During the down stroke, the relative increment of NACA0012 airfoil continuously increases and obtains the maximum value of −2.47% at the angle of attack of 7.5°. Then, it keeps a slow decrease until the end. The relative increment of S809 airfoil rapidly decreases first and obtains the value of −3.5% at the angle of attack of about 12°. In the middle section, a fluctuation is shown, and the cause is revealed above. From 6.5° to 2.5°, the relative increment keeps decreasing. Obviously, areas bounded by the two relative increment loops; in Case 2, it is larger than those in Case 1, which illustrates the grater difference of the former between up and down strokes.
In
Figure 10 and
Figure 11, for Case 3, during the up stroke, relative increments of two airfoils are both smaller than 1% in most sections, except for that of S809 airfoil at angles of attack smaller than 4°. A maximum value of −3.02% is obtained at the minimum angle of attack. During the down stroke, the increment of NACA0012 airfoil has a mean value of −1.9%, and the maximum value of −2.32% is obtained at the angle of 7.6°. For S809 airfoil, the relative increment rapidly increases from −1.37% at 13.5° to −6% at 12.3°. Then, it increases again with a small slope. The maximum value, which is about −7%, appears at a middle angle of attack. In addition, areas bounded by the relative increment loops of two airfoils further increases, and it is greatly obvious for S809 airfoil. Therefore, a larger reduced frequency can increase the range of the large aerodynamic performance loss and the difference between two airfoils, especially during the down stroke.
3.2. Hysteresis Loops at Different Mean Angles of Attack
In the previous section, the comparison is presented at a mean angle of attack of 8°. Another two mean angles of attack are adopted to further reveal their difference. The default particle mass concentration is also 1% and the more details are listed in
Table 3. Then, the computed results of these three cases are discussed, respectively.
In
Figure 12a, for Case 4, the loop of NACA0012 airfoil is hardly changed by particles, due to its insensitivity during both up and down strokes. However, in
Figure 12b, the loops of S809 airfoil during the beginning of the down stroke have apparently smaller lift coefficients than those during the end of the up stroke, which is caused by the light stall. Although particles have an obvious change in this region, it is still too small to be paid attention to. Therefore, the effect of particles on the lift coefficient is slight and negligible at the default particle mass concentration in the whole loop.
For two airfoils, the angle of attack, at which the maximum lift coefficient is obtained, is unchanged by particles. In order to ascertain this phenomenon, two larger particle mass concentrations of 3% and 5% are adopted, and the results are listed in
Table 4. It is found that particles do not change the angle at which the maximum lift coefficient is obtained, even though it is under the effect of light stall at moderate and high reduced frequencies.
The comparison for Case 2 was presented in
Figure 8 and
Figure 9. In
Table 5, for NACA0012 airfoil, all angles of attack, at which the maximum lift coefficient is obtained, were 13.5°, while for S809 airfoil, the value decreases with the mass concentration. This may illustrate that the angle of attack, at which the maximum lift coefficient angle is obtained, can be changed by particles in the region, which suffers the effect of deep stall.
Finally, as shown in
Figure 13 for Case 5, two airfoils both suffer the effect of deep stall and have different degradations of the lift coefficient near the maximum angle of attack. In
Figure 14, during the up stroke, the relative increment of NACA0012 airfoil is smaller than −1.25% at angles of attack less than 15° and increases to −3.01% at 16.5°. With the angle of attack decreasing, it increases again at a greater speed rate. Then, the maximum increment of −7.02% is obtained at 13.9°. In the following region, the relative increment linearly decreases until the angle of attack reaches to 6.2°. For S809 airfoil, the relative increment is smaller than −1.5% at angles of attack less than 12° and rapidly increases at following angles of attack. The first peak value of −7.23% appears at 15.2° and it decreases to −3% at 16.5°. During the down stroke, the relative increment increases again, and the second peak value is −7.05% at 9.5°. Then, the decrease runs through the rest region. In brief, particles have large effect on the lift coefficient of both airfoils during the down stroke. However, during the second half of the up stroke, particles also have a large effect on the lift coefficient of S809 airfoil.
Table 6 shows angles of attack, at which the maximum lift coefficient is obtained, of two airfoils at different particle concentrations. They both decrease with the particle mass concentration. In summary, particles can change the angle of attack, at which the maximum lift coefficient is obtained, in the region under the effect of deep stall.