This study experimentally and numerically investigates the effect of application of curvilinear element blades to fully-shrouded centrifugal compressor impeller on the performance of centrifugal compressor stage. Design suction flow coefficient of compressor stage investigated in this study is 0.125. The design guidelines for the curvilinear element blades which had been previously developed was applied to line element blades of a reference conventional impeller and a new centrifugal compressor impeller with curvilinear element blades was designed. Numerical calculations and performance tests of two centrifugal compressor stages with the conventional impeller and the new one were conducted to investigate the effectiveness of application of the curvilinear element blades and compare the inner flowfield in details. Despite 0.5% deterioration of the impeller efficiency, it was confirmed from the performance test results that the compressor stage with the new impeller achieved 1.7% higher stage efficiency at the design point than that with the conventional one. Moreover, it was confirmed that the compressor stage with the new impeller achieved almost the same off-design performance as that of the conventional stage. From results of the numerical calculations and the experiments, it is considered that this efficiency improvement of the new stage was achieved by suppression of the secondary flows in the impeller due to application of negative tangential lean. The suppression of the secondary flows in the impeller achieved uniformalized flow distribution at the impeller outlet and increased the static pressure recovery coefficient in the vaneless diffuser. As a result, it is thought that the total pressure loss was reduced downstream of the vaneless diffuser outlet in the new stage.

The effects of a return channel with splitter vanes on the performance of a multistage centrifugal compressor were investigated. As a preliminary study, the optimum location of the splitter vanes was numerically examined with the aim of achieving high overall efficiency. The results indicated that the optimum location was the 30% of the normalized pitchwise distance from the suction side of the main vane, with the leading-edge located at a radius ratio to the main vane trailing-edge of 1.77. To investigate the effects of the return channel with and without the optimum splitter vanes on the overall performance, performance tests were carried out using a one-and-half-stage test rig. Three pre-swirl vanes, whose vane angles from the tangential direction at the trailing-edge were 20, 30 and 40° were used to simulate three operating conditions with low, design and high flow coefficients, respectively. The design flow coefficient of the downstream impeller was 0.073 and the peripheral Mach number was 0.87. The test results showed that the return channel with the optimum splitter vanes achieved 11.8% higher overall efficiency at the high flow coefficient with respect to the case without the splitter vanes while maintaining the same efficiency at both low and design flow coefficients. The return channel with the optimum splitter vanes was concluded to be effective for improving the efficiency of a multistage centrifugal compressor.

Experimental and numerical studies were performed to investigate influences of the return channel flow on the surge margin of a multistage centrifugal compressor. Two return channels, which were named RCH-A and RCH-B, were designed and evaluated in a two-stage centrifugal compressor. The measured performance of the compressor suggested that the surge margin of this compressor was dominated by the operating limit of the second stage and that the surge margin of RCH-B was 5% larger than that of RCH-A. The outlet flow of RCH-A and RCH-B swirled in a counter-rotating direction near the shroud region, and the flow angle at the outlet of RCH-A was larger than that of RCH-B. CFD was conducted to investigate the internal flow in the return channel. The CFD results of both RCH-A and RCH-B showed that the flow separation occurred on the suction surface of the return vanes near the operating limit. This separation induced the velocity difference between the suction and pressure sides, and the swirl flow in the counter-rotating direction was generated by this velocity difference. The swirl flow in the counter-rotating direction increased the blade loading of the second stage impeller at the operating limit. It was considered that the blade loading of RCH-B was lower than that of RCH-A at the operating limit, because the swirl flow in the counter-rotating direction of RCH-B was weaker than that of RCH-A. Therefore, the surge margin of the second impeller with RCH-B seemed to be larger than that with RCH-A. It was conclude from the experimental and numerical results that the locally swirl flow in the counter-rotating direction at the outlet of the return channels near the shroud side influenced the surge margins of the downstream impeller and the multi-stage centrifugal compressor.

The influence of the boss-tip radius ratio of the impeller on the aerodynamic performance of a multistage centrifugal compressor were numerically and experimentally investigated. Three impellers, which boss-tip radius ratios were 0.28, 0.34, and 0.40, were developed for a design suction flow coefficient of 0.125. The inlet radius at the shroud of each impeller was designed to provide the minimum inlet relative velocity, and the exit width was also designed to keep the inlet to exit relative velocity ratio. The numerical and experimental results suggested that the impeller pressure-rise decreased as the boss-tip radius ratio increased. The decrease of the impeller efficiency due to the increase of the boss-tip radius ratio could be minimized by minimizing the inlet relative velocity at the shroud side and by keeping the inlet to exit relative velocity ratio constant in the tested cases. The ribbed diffuser seemed to be useful to improve the performance of compressor stage with the high boss-tip radius ratio impeller, although further investigations will be needed to clarify the change in the flow fields and the loss generation mechanism in the ribbed diffuser with the high boss-tip radius ratio impeller.

This study numerically and experimentally examines the effects of applying curvilinear element blades to fully-shrouded centrifugal impellers on the performance of the centrifugal compressor stages. The design suction coefficient of the target impellers was 0.073. Our previous study confirmed that the application of curvilinear element blades could improve the stage efficiency of similar types of centrifugal compressors. However, a detailed explanation of the relation between the stall margin and the application of the curvilinear element blades remains to be given. The purpose of this study is to investigate the effects of using these blades on the impeller flow field and the stall margin in further detail. The curvilinear element blades we developed for centrifugal turbomachinery were defined by the coordinate transformations between a revolutionary flow-coordinate system and a cylindrical coordinate system. All the blade sections in the transferred cylindrical coordinate system were moved and stacked spanwise in accordance with the given “lean profile,” which meant the spanwise distribution profile of movement of the blade sections, to form a new leaned blade surface. The effects of the curvilinear element blades on the impeller flowfield were investigated by conducting numerical simulations using this method. We next considered the optimum design guidelines for impellers with curvilinear element blades. Then we designed a new impeller using these design guidelines and evaluated the performance improvement of a new compressor stage by conducting numerical simulations. As mentioned in several papers, we numerically confirmed that curvilinear element blades with a negative tangential lean profile improved the velocity distribution and stage efficiency because they help to suppress the secondary flows in the impeller. The negative tangential lean mentioned in this paper represents the lean profile in which the blade hub end leans forward in the direction of the impeller rotation compared to the blade shroud end. At the same time, we also found that the stall margin of these impellers deteriorated due to the increase in relative velocity deceleration near the suction surface of the shroud in the forward part of the impeller. Therefore, we propose new design guidelines for impellers with the curvilinear element blades by applying a negative tangential lean to line element blades in which the blade loading of the shroud side in the forward part of the impeller is reduced. We confirmed from the numerical simulation results that the performance of the new compressor stage improved compared to that in the corresponding conventional one. The new design guidelines for the curvilinear element blades were experimentally verified by comparing the performance of the new compressor stage with the corresponding conventional one. The measured efficiency of the new compressor stage was 2.4 % higher than that of the conventional stage with the stall margin kept comparable. A comparison of the measured velocity distributions at the impeller exit showed that the velocity distribution of the new impeller was much more uniform than that of the conventional one.

This paper describes the optimal design and experimental verification of centrifugal compressors with leaned curvilinear element blades. The design targets were a fully shrouded centrifugal impeller and a low-solidity vaned diffuser. First, a new method of defining the curvilinear element blade was developed for centrifugal turbomachinery using coordinate transformations between a revolutionary flow-surface coordinate system and a cylindrical coordinate system. All the blade sections in the transferred cylindrical coordinate system were moved and stacked spanwise to form a new leaned blade surface. The inverse transformation results in a curvilinear element blade in the original coordinate system. The direction of movement of the blade section could be any of the existing definitions such as sweep, dihedral, or tangential lean. For simplicity, we have defined a “lean profile” as a general expression for the spanwise distribution profile of movement of the blade sections. Model compressors with curvilinear element blades were then designed at a suction flow coefficient of 0.073. Optimal lean profiles for the impeller and vaned diffuser that maximized adiabatic efficiency and uniformity of outflow were explored using a multi-objective genetic algorithm, Kriging surrogate model, and steady Reynolds-averaged Navier Stokes simulations. We chose efficiency-weighted solutions since the efficiency and uniformity had a positive correlation. A sensitivity analysis showed that tangential leans near the endwalls are keys for improving the impeller’s efficiency. The chosen optimal impeller had a concave blade suction surface and a concave leading edge. Although clear patterns in geometrical features for optimal diffuser vanes could not be captured, we found that the dihedral profile had a predominant effect on efficiency. The model compressors was experimentally measured and compared with traditional compressors as to their aerodynamic performance. The experimental apparatus was composed of a suction nozzle, impeller, diffuser, and return channel. The results demonstrated that the models’ adiabatic efficiency was higher by 1.2–1.4% at the design point. Although the stall margins slightly decreased (by 0.7%) in the experiments, the surge margins expanded (by 3.7–6.7%).

Design parameters for a suction channel of process centrifugal compressors were investigated, and an optimizing method to improve efficiency by using the new design parameters was proposed. Both pressure loss and circumferential flow distortion in the suction channel were evaluated by using computational fluid dynamics (CFD). The main dimensions, which had a large influence on pressure loss and circumferential flow distortion, were identified by using design of experiments (DOE). Next, the passage sectional area ratios A c /A e , A e /A s , and A c /A s were found to be the dominant design parameters for the pressure loss and circumferential flow distortion, where A c , A e and A s are passage sectional areas for the casing upstream side, casing entrance and impeller eye, respectively. Then the shape of the suction channel was optimized using A c /A e , A e /A s , and A c /A s . Finally, to evaluate the improvement effect of optimizing the values of A c /A e , A e /A s , and A c /A s on compressor stage performance, a base suction channel and an optimized type of suction channel were manufactured and tested. The design suction flow coefficient was 0.1 and the peripheral Mach number was 0.78. Test results showed that the optimized suction channel achieved 3.8% higher stage efficiency than the base one while maintaining the overall operating range from surge to choke. The method for optimizing suction channels by using the three described design parameters was concluded to be very effective for improving the stage efficiency.

The authors previously found that compressor stage efficiency in a high specific speed range was significantly improved by employing an increased relative velocity diffusion ratio coupled with a high backsweep angle. In spite of such a high relative velocity diffusion ratio, the same surge margin as with a conventional design could be achieved by using a special front loading distribution with a lightly loaded inducer. In the present study, the blade loading distribution was further optimized in order to achieve a larger surge margin than previously. Four types of fully shrouded impellers were designed, manufactured and tested to evaluate the effects of blade loading, backsweep angle and relative velocity diffusion ratio on compressor performance. The design suction flow coefficient was 0.125 and the machine Mach number was 0.87. Test results showed that the developed impeller achieved 3.8% higher stage efficiency and 11% larger surge margin than the conventional design without reducing the pressure coefficient and choke margin. It was concluded that aft loading coupled with a high degree of reaction was a very effective way to improve surge margin as well as stage efficiency. Stator matching was also investigated by changing the design incidence angle which was shown to have little influence on surge margin in the present test results.

Performance improvement of 3D impellers in a high specific speed range was investigated using computational fluid dynamics (CFD) analyses and experimental tests. In order to reduce the loss production within the stator passages, the backsweep angle of the impellers was increased. At the same time, the inlet-to-exit relative velocity diffusion ratio was also increased by increasing impeller exit width to prevent the reduction in pressure ratio. Moreover, the blade loading distribution at the impeller shroud side was optimized to suppress the surge margin reduction caused by the increased relative velocity diffusion ratio. Five types of unshrouded impellers were designed, manufactured and tested to evaluate the effects of blade loading, backsweep angle and relative velocity diffusion ratio on compressor performance. The design suction flow coefficient was 0.125 and the machine Mach number was 0.87. Test results showed that compressor stage efficiency was increased by 5% compared to the base design without reducing the pressure coefficient and surge margin. It was concluded that an increased relative velocity diffusion ratio coupled with large backsweep angle was a very effective way to improve compressor stage efficiency. An appropriate blade loading distribution was also important in order to achieve a wide operating range as well as high efficiency.

The improvement of efficiency and operating range by optimizing blade-loading distribution of three-dimensional (3D) centrifugal compressor impellers is investigated using computational fluid dynamics (CFD) analyses and performance tests. The design points of suction flow coefficients investigated in the present study were 0.05 and 0.073. Two design approaches for 3D impeller were employed: a conventional method and a newly developed one. In order to achieve higher efficiency and wider operating range, the blade loading and relative velocity distribution were optimized in the new design procedure. In addition, to clarify the performance improvement of 3D impellers against current two-dimensional (2D) ones, both performance characteristics were compared. The test results showed the efficiencies of the newly designed 3D impellers were increased by about 0.5–1.5% in comparison with those of the conventional impellers, while the operating ranges of both were almost the same. Further, the efficiencies of the newly designed 3D impellers increased by about 3% in comparison with those of the 2D impellers at both design points. At the same time, the operating ranges of the former impellers were about 2.1–2.8 times as wide as those of the latter.