3.1. Analysis of Wear Results
After the centrifugal pump has undergone 48 h of the solid–liquid two-phase flow abrasion experiment, the internal flow passage of the centrifugal pump has different degrees of wear. Analyzing the results of the wear experiment, it was found that the most severely worn parts in the impeller flow passages are the hub wall and the leading edge of the blade. Under different mass concentration conditions, the hub wall wear is shown in
Figure 5.
Based on the wall wear results obtained from the experiment and simulation shown in
Figure 5, it was obvious that when the centrifugal pump performed the solid–liquid mixed transportation of different concentrations, the wall of the hub near the exit of the impeller flow passage first produced wear, and the wear at this place caused the paint surface to disappear first, exposing the metal surface of the hub wall. It can be found that when the particle mass concentration increased from 2% to 3%, the paint on the wall of the hub disappeared gradually, and the amount of wear also increased, which indicated that the wear range and the wear degree were closely related to the particle concentration. This phenomenon occurs in both simulations and experiments. With further observation of the wear on the wall of the hub, we found that the wear surface had corrugated stripes, which were very similar to the striped wear shown on the wall of the elbow [
15].
In order to study the cause of the wear of the hub, the solid–liquid two-phase flow inside the region was analyzed. Since the impeller used in this study was a double-curvature blade, in order to be able to show its internal flow characteristics, the impeller’s rotating flow area was expanded into a blade-to-blade region. The diagrams of the blade-to-blade region shown below in
Figure 6 at different positions were numbered by the span value, from 0 to 1, representing the blade tip to the root; that is, the smaller the span value was, the closer the position was to the blade tip.
The effect of the presence of solid particles on the flow was studied to investigate the wall wear law by analyzing the flow inside the impeller when the centrifugal pump conveyed water and the solid–liquid two-phase flow.
Figure 7 shows the flow and velocity diagrams on the impeller grid expansion surface (Span = 0.9) near the hub area.
As can be seen from
Figure 7, in the liquid flow condition, the streamline in the area near the hub wall was smooth, the velocity distribution was relatively uniform, and there was no obvious vortex. However, in the solid–liquid two-phase flow condition, there were several low-speed areas in the impeller flow passage, and the flow velocity near the outlet increased with the appearance of the vortex, which corresponded to the earliest disappearance of the hub outlet position in the wear experiment. That is, the fast-flowing fluid entrained the solid particles to rotate and rub at that location, and then wear first occurred in this part. The greater the concentration was, the greater the number of particles was, and the greater the degree of wear and the range of wear were.
Combining the particle motion vector diagram could further reveal the cause of the wheel hub wall wear, as shown in
Figure 8.
It can be seen from
Figure 8 that after entering the impeller flow area through the inlet pipe, a large number of particles did not collide with any surface to decelerate, but rather directly impacted the wall surface of the impeller hub. Therefore, severe wear occurred at this position first.
It can also be observed from
Figure 8 that after the particles entered the impeller flow channel, the wall of the impeller hub was impacted by the particles, and the leading edge of the blade was directly impacted by a large number of particles, as shown in
Figure 9. Therefore, the blade leading edge of the impeller was also severely worn. The experimental and calculated wear results are shown in
Figure 10.
It can be clearly seen from
Figure 9 that when the particles entered the impeller flow area, the leading edge of the blade was directly impacted by a large number of particles. It can be seen from
Figure 10 that after 48 h of abrasion testing at 1% concentration, the paint peeled off on part of the leading edge of blade, while the remaining part of the paint surface remained intact. There were many small pits on the leading edge of the blade. The wear condition was basically consistent with the results obtained by the simulation calculation, and the wear type was mainly impact wear. As the concentration increased, the wear area increased and the degree of wear continued to deepen, but the types of wear were essentially the same. For the case of 3% concentration, the leading edge of the blade was impacted by particles and it disappeared, and at the same time, dents appeared in the blade leading edge. This might have been caused by the continuous impact of particles. At 5% concentration, the leading edge of the blade became smoother, the paint on the blade leading edge was completely peeled off, and a certain degree of distortion appeared.
As the particle concentration increases, it could be seen that the wear degree in the leading edge of the blade area increases, while the wear area expands accordingly. After the paint disappeared completely, pits gradually appeared on the leading edge of the blade. As the concentration further increased, the leading edge of blade gradually became smooth and rounded, and even deformed.
When the liquid phase drove the solid particles to continue to move, they entered the rotating impeller channel through the blade leading edge of the impeller and the hub, and then the particles and fluid rotated under the action of the impeller. The wear of the blade pressure surface is shown in
Figure 11.
It can be seen from
Figure 11 that the indicating paint on the pressure surface was only retained for the working condition of 1% concentration, and the indicating paint surface of the pressure surface disappeared completely for the conditions of other concentrations. At the same time, the pressure surface wear trend in the experiment was consistent with the simulation.
The wear of the pressure surface appeared from the root and tail of the pressure surface and then gradually spread to the entire pressure surface. It could be seen that the paint on the pressure surface disappeared completely in the condition of 3% concentration, and the pressure surface was worn smoother for 5% than other concentrations.
From the leading edge of the blade to the tail of the blade, we took a line in the middle of the blades and divided it with 12 points, and named the monitoring points from 1 to 12 in the direction of the arrow, and the thickness loss of the impeller was measured at these monitor points.
Figure 12 shows the monitoring point on a blade and the mark of the impeller blade.
Figure 13 shows the wall thickness loss at 1% concentration in the experiment and the simulation.
In
Figure 13, the thickness loss was measured in the experiment and simulation. In order to be able to compare the two situations, the thickness loss rate was calculated by dividing the thickness loss by time, respectively.
It can be seen from
Figure 13 that both in the experiment and simulation, at the measuring point near the blade leading edge (monitor point 1), the three blades all had serious thickness loss. When the measuring point is gradually moving away from the blade leading edge (at the monitor points 2 and 3), the thickness increases to a certain extent due to particle impact in the experiment. As the measuring point moves to the root of the impeller, the thickness loss gradually becomes serious.
The location of the largest thickness loss in the experimental wear is at the tail of the blade. The wear value of the three blades at the monitoring point 12 has an average increase of 215.53% compared to the wear value of the monitoring point 1.
It can be seen from
Figure 13 that the thickness loss rate trend of the simulation is consistent with the experimental data, and the most severely worn point is 12 at the tail of the blades. For the blades b and c, the wear of the monitoring point 12 and the wear of the monitoring point 1 increased by an average of 232.35%, which is consistent with the experimental results.
The thickness loss in the experiment situation was greater than that in the simulation. The experimental value has a large fluctuation amplitude, which is due to the change of the impeller surface due to wear, and there are preliminary slight wear streaks.
In order to better explain the occurrence of wear, the streamline and the velocity distribution of the flow field were analyzed with the cascade diagram.
It can be seen from
Figure 14 that the vortex and the corresponding flow chaotic position at the same concentration basically appeared in the middle area of the blade pressure surface, and the flow chaotic vortex always existed no matter how the position of the cascade section changed. Taking the 1% particle concentration as an example, it could be seen that when the span value near the blade tip was 0.1, the streamline was relatively smooth, the streamline vortex was mainly concentrated in the middle of the blade pressure surface and the position of the leading edge of the blade, and the streamlines in the three flow passages of ABC were similar. With the continuous increase of the span value (that is, corresponding to different sections from the tip to the root of the blade), it could be clearly seen that for the flow lines in each flow passage, a certain degree of difference gradually appeared, but the phenomenon of vortex near the pressure surface still existed. The particles at the pressure surface might have been affected by the flow field and repeatedly impacted the pressure surface. At the same time, the circumferential velocity of the pressure surface had a certain difference with the particles too, so the wear had to start from the end of the pressure surface with the highest circumferential velocity and gradually spread to the entire pressure surface.
3.2. Internal Flow Field Analysis
When the centrifugal pump performed solid–liquid two-phase transportation, the presence of particles had a great impact on the flow field, and the change of the flow field in turn affected the movement and distribution of the particles, which in turn affected the mixed transport performance and wear. The flow field of the two-phase conveying condition and the flow field of the pure liquid conveying condition were compared in order to explore the influence of the particles on the flow field.
It can be seen from
Figure 15 that in the liquid flow condition, the streamlines in the flow field appeared to be very smooth, and there was no particularly obvious vortex. At the same time, the cloud atlas of the flow field changed uniformly in this case. In the two-phase flow condition, it was obvious that there was a vortex near the pressure surface in the middle of the passage, and the closer the position was, the larger the vortex was. At the same time, when the span value was 0.9, which was the closest to the bottom of the blade, another vortex appeared near the leading edge of the blade. It could be seen from the velocity distribution of the flow field that as the section position gradually moved from the top to the bottom of the blade, the velocity of the flow field gradually decreased, the vortex range became larger, and another vortex even appeared near the bottom of the blade.
In order to further study the influence of the particles on the energy loss of the flow field, the energy gradient theory [
16,
17] was introduced for flow analysis.
The expression of the energy gradient,
K [
18], in the centrifugal pump was:
In this expression, K is defined as the energy gradient function, is the velocity, E is the total mechanical energy, is defined as the normal gradient of the total mechanical energy in the streamline, and is defined as the flow gradient of the total mechanical energy. In this fraction, the denominator term is the flow gradient of the work carried out by the viscous shear stress in the streamline direction, which can be further expressed as the sum of the viscous friction loss of the total mechanical energy in the streamline direction and the energy dissipation function, .
The energy dissipation function,
, can be defined as:
The energy gradient theory could be used to predict the increase or loss of the flow field energy and the K value could be approximated as the ratio of the energy increase to the energy loss. To make it easier to observe, the logarithm lgK of the K function was used for the display. It was defined that lgK in the range of −0.3 to 0.3 meant that the local energy loss was equal to the energy increase. When lgK ≥ 0.3, the energy increase was dominant, and lgK ≤ −0.3 indicated that the energy loss was dominant. The energy change trend of the flow field reflected the corresponding energy change trend of the particles, which could be used to predict particle distribution and the movement to a certain extent. At the same time, the flow field tended to flow from a position with higher energy to a position with relatively low energy. The field tended to flow in a more complicated manner in areas where the energy increase was dominant.
Due to the different positions of the flow passages at the current moment and the direction of gravity, the particle distribution in each flow passage and each span had to be different. In
Figure 16, near the top of the blade, when the span values were 0.1 and 0.3, the high-concentration particle volume fraction appeared at the entrance of the C flow passage. This was because the particles were gradually sinking, they were affected by gravity before entering the front of the impeller zone, and the inlet of the C flow channel was just below the flow passage at that moment. At that time, a large number of particles entered the C flow channel. In the particle distribution diagram, it can also be seen that as the span value increased, the particles were concentrated at the entrance from the beginning and they gradually spread to the entire flow passage. At the same time, the particle volume fraction in front of the pressure of the B channel increased with the increase of the cross-sectional span value, and most of the particles were free near the bottom of the blade pressure surface, which led to the wear of the blade pressure surface, as shown in
Figure 11. When the span value was 0.9, close to the wall of the impeller hub, there were a large number of particles in the wall and a large amount of the volume was occupied. This was also the reason that the wear of the hub could spread to the entire wall.
It could be seen from the distribution of lg
K in
Figure 16 that when the span value was 0.1, there was a large range of high-energy areas in each flow passage, mainly concentrated in the center of the suction surface and close to the impeller outlet area, and there was almost no position where the energy was weakened. It could be known from the particle distribution that the particles had not completely entered the flow passage when the span value was 0.1. At other span values, areas where the energy was weakened gradually appeared, and particles also began to enter the flow passage. This indicated that the main reason for the energy weakening of the flow field was the existence of particles, and the flow field drove the movement of particles by transferring energy to the particles.
Combined with the current concentration situation and the distribution of lgK on the sections with different span values, it could be known that when the span value was 0.1, the flow field had a large range of high-energy regions. As the span value continued to rise (that is, corresponding to the section from the tip to the root of the blade), it could be clearly observed that the high-energy area gradually decreased, while the energy balance area occupies most of the flow passage. According to the characteristics that fluids tend to flow from high-energy regions to low-energy regions, it could be considered that in the vertical scale, the flow field continuously flowed from the tip to the root of the blade, which also made most of the particles receive energy transfer in the impeller flow passage. Thereby, the particles continuously accumulated in the root area of the blade, and since most of the energy of the flow field was in the equilibrium range near the root of the blade, the particle concentration in that area was relatively higher.
In order to understand the influence between the particles and the flow field, the lgK distributions of both conditions were compared. When the span values were 0.1 and 0.3, it could be seen that the energy increase was dominant (lgK ≥ 0.3) and it mainly appeared in the middle of the flow passage, and most of the energy increase area had a certain distance from the pressure surface and were close to the suction surface. This was also similar to the condition of the water. It could be known that due to the presence of particles in the current cascade section, the energy transferred to the particles by the flow field was relatively small, and the energy gradient distribution in the three flow passages was relatively similar.
However, in the area where the span value was 0.5, significant differences began to appear for the conditions of 0% and 1% mass concentrations. The exit position of the flow passage was in the area of reduced energy for the condition of clear water, but when there were particles, either it completely disappeared, or the energy increase was dominant. This might have been because the particles further transferred their own energy to the flow field at the current position.
In the section near the root of the blade (span values of 0.7 and 0.9), the difference between 0% and 1% particles was more obvious, especially when the span value was 0.9. The entire flow passage basically had the energy increase dominate for the condition of water, but when particles existed, the energy increase dominated the area in the flow passage that appeared near the suction surface.
By analyzing the overall flow characteristics of the flow passage, it could be found that compared with the K value in the flow passage for the condition of clear water transportation, the K value of the particle aggregation area during two-phase transportation was relatively smaller. The area near the root of the blade with the most particles (the span value was 0.9) had the most significant decrease in the K value. Since the K value could indicate the local energy situation, a small K value meant that this was a low-energy area, and the flow was relatively stable, so the particles aggregated there.
The flow loss in the impeller channel mainly included the boundary layer flow loss and the main channel flow loss. It could be found that a certain number of particles near the wall could disturb the boundary layer during solid–liquid two-phase transportation, reducing its thickness and thereby reducing the boundary layer flow loss. However, when the concentration was too high, the presence of particles caused the overall flow field disturbance, which led to an increase of the main channel flow loss. Therefore, combined with the head curve of the centrifugal pump in
Figure 4, it could be found that at low concentrations, the reduction of the boundary layer flow loss was dominant, so the head change of the solid–liquid two-phase flow was small. After the increase in concentration, the decrease in the boundary layer flow loss was no longer sufficient to compensate for the increase in the main channel flow loss, which led to an increase in the flow loss in the impeller flow channel, so the head was greatly reduced.