Quasi-Dynamic Experimental Investigation of the Internal Flow Field in the Stationary Components of a High-Flow-Coefficient Centrifugal Compressor

Inside the heart of modern industry, from the massive turbines that power chemical plants to the compact engines that drive aircraft, lies a machine that seems deceptively simple: the centrifugal compressor. Its job is to squeeze air, increasing its pressure to make it useful for combustion or storage. While the spinning blades that do the heavy lifting have been perfected over decades, reaching near-perfect efficiency, the stationary parts that guide the air afterward have become the next frontier for improvement. These stationary components—a series of curved passages and channels that sit still while the air rushes through them—account for a small but critical slice of energy loss. In high-performance machines designed to move large volumes of air, these stationary parts are where engineers now look to squeeze out extra efficiency. The challenge has always been that these parts are hidden deep inside thick metal casings, making it nearly impossible to see how the air actually moves inside them while the machine is running.

To solve this puzzle, a team of researchers at the TaiHang National Laboratory in China built a special experimental setup that acts like a time machine for airflow. Instead of spinning the main blades at thousands of revolutions per minute, which would make it too dangerous and difficult to insert measuring tools, they replaced the spinning impeller with a set of stationary guide vanes. By carefully rotating these vanes in tiny, precise steps, they simulated the effect of the spinning blades passing by, effectively recreating the complex, shifting patterns of air that would occur in a real, running engine. This "quasi-dynamic" approach allowed them to use a high-precision probe, shaped like a small five-hole needle, to map the air's speed, pressure, and direction at four distinct locations deep inside the machine's stationary passages. They were looking for the "wake"—the trail of slower, turbulent air left behind by the blades—and wanted to see exactly how it changed as it traveled through the diffuser, the U-shaped bend, and the return channel.

The researchers focused on a specific type of high-flow compressor, one designed to move air at a rate of 0.2 flow coefficient with a machine Mach number of 0.8. They set up their test rig inside a large wind tunnel and began by rotating their stationary guide vanes through one full blade pitch, stopping at eight different angles to take measurements. At the very first measurement point, just after the air left the guide vanes, they found a clear, repeating pattern of high and low-pressure zones, mirroring the shape of the blades. As the air moved into the next section, a wide, open space called the vaneless diffuser, the researchers watched this pattern begin to blur. The air mixed together, smoothing out the sharp edges of the wake, but the distinct peaks and valleys of pressure were still clearly visible, shifting in perfect sync with the angle of the vanes.

The most dramatic transformation happened when the air entered the U-bend, a 180-degree turn that redirects the flow toward the next stage. Here, the researchers observed a powerful homogenizing effect. The sharp, uneven patches of pressure that had survived the diffuser were almost entirely washed out by the turn. The U-bend acted like a giant mixer, driven by the physics of the curve itself, which forced the air from the inner and outer walls to swirl and exchange places. By the time the air exited the bend, the researchers found that the unevenness of the pressure had dropped by about 90 percent. The chaotic, shifting wake that had been so prominent just moments before had been smoothed into a nearly uniform flow. This was a crucial discovery, proving that the bend does more than just change the direction of the air; it fundamentally cleans up the flow quality before it enters the next set of components.

However, the story did not end with a perfectly smooth flow. As the air moved past the bend and approached the return channel vanes, a new phenomenon appeared. Even though the original wake from the blades had vanished, a new, smaller ripple in the pressure emerged. This ripple was not caused by the blades at all, but by the stationary vanes themselves. The researchers discovered that the shape and position of these vanes created a subtle, invisible pressure field that reached backward, influencing the air just before it arrived. This "potential-flow feedback" meant that the air felt the presence of the vanes before it actually touched them, creating a predictable, repeating pattern that matched the spacing of the vanes. It was a reminder that in fluid dynamics, the parts of a machine do not just act on the air passing through them; they also whisper to the air approaching them.

By the time the air reached the final exit of the stationary section, the original wake from the blades was completely gone. The flow was perfectly uniform as it circled around the machine, but it was not uniform from top to bottom. The air near the top and bottom walls moved at different speeds and angles compared to the air in the middle, creating a vertical gradient that could affect the performance of the next stage. Perhaps the most surprising finding of the entire study concerned the energy loss. The researchers expected that the position of the wake might change how much energy the machine lost, but they found that the total loss remained constant regardless of where the wake was. Instead, the losses simply shifted between the U-bend and the return channel. When the bend lost more energy, the return channel lost less, and vice versa. They traded off perfectly, keeping the total cost of the journey the same.

This study provides a clear, detailed map of how air behaves inside the hidden corridors of a high-performance compressor. It confirms that while the spinning blades create the initial disturbance, the stationary components are powerful agents of change, capable of smoothing out chaos and redistributing energy in ways that were previously difficult to measure. The researchers have shown that the total energy loss of the return passage is determined by the average conditions of the air entering it, not by the specific, shifting position of the wake. This insight offers a simplified rule for engineers designing these machines: they can focus on the overall flow conditions rather than trying to track every individual ripple of air. The work also highlights that while the U-bend is excellent at mixing the flow, it leaves behind a vertical imbalance that will need to be addressed in future designs to ensure the next stage of compression runs as smoothly as possible.