No abstract

The temperature depression of liquids due to latent heat of vaporization causes vapor pressure depression and suppresses cavity growth. This phenomenon is called ''thermodynamic effect of cavitation.'' This effect is especially significant in cryogenic fluids such as LOX and LH 2 . Due to this effect, the performance of hydraulic equipment for cryogenic fluids, such as turbopumps of rocket engines, is not as bad as predicted. In this paper, the size of the cavity in cryogenic fluid is estimated numerically taking the thermodynamic effect of cavitation into consideration. A cavity is assumed to be a sheet cavity. From the results, the effects of the properties of liquid and Reynolds number on the thermodynamic effect of cavitation are investigated.

In order to improve overall performance of a turbomachinery, it is important to reduce losses of stationary flow passages, such as diffusers and return channels, as well as impellers. For multi-stage pumps, to achieve high uniformity of the inlet flow of the latter impeller can prevent degradation of subsequent performance. A two stages high pressure pump which was developed by the authors applied vaned diffuser sections located downstream of the first stage impeller and second stage impeller. The vaned diffusers can be replaced to fit the required specification of operating condition. This design is able to help pumps share the casings so that customers can purchase our products at a low cost. However, the loss of the first stage diffuser section and crossover downstream of the first stage diffuser to the second impeller is large, so that it has a negative impact on overall performance of the pump. It was difficult to reduce loss of the both sections by conventional way such as trial and error approach by modifying geometrical parameters. If we try to reduce the loss of the diffuser section, loss of the crossover passage increases. We therefore applied a technology called Adjoint method to the design optimization of the diffuser and crossover sections of the pump. The adjoint method has recently been put to practical use. By using this method it is possible to obtain a complex three-dimensional shape for realizing an optimum flow field in a very short time while maintaining a high degree of freedom. In this study two objectives were specified in the design optimization of the stationery sections; one is to increase the uniformity of the flow field at the inlet of the second stage impeller and the other is to minimize the loss of the first stage diffuser and crossover passage. In the optimization process, the change of sensitivity of the geometry deformation to the weights of objective functions was analyzed by sensitivity maps. The sensitivity maps show directions of the geometry modification depending on the objectives. The geometry in this design was optimized with 50% - 50% weighting ratio between flow uniformity and pressure loss. The loss of the vaned diffuser and crossover passage, velocity uniformity at the outlet of the crossover passage were verified with Computational Fluid Dynamics (CFD). The improvement of pump stage performance was expected by the optimization. This optimization design procedure applying Adjoint method is effective to improve overall performance of a two stages high pressure pump which has complex three-dimensional flow passages.

The following study describes the optimization design procedure of a double-suction pump. BASELINE pump is designed as inlet nozzle diameter 800 mm and impeller outlet diameter 740 mm. Each component of a BASELINE pump, impeller configurations, discharge volute, and the suction casing were determined by DOE (Design of Experiments) and sensitivity analysis. However, finite selected design parameters for each component are mostly restricted to the free surface design of the pump casing. In this study, the optimization method approach along with steady Computational Fluid Dynamics (CFD) is introduced to achieve the high efficiency request of a double-suction pump. To investigate the matching optimization of the impeller and discharge volute at design point, the full parametric geometry of discharge volute was developed referred to the BASELINE shape and Multi-Objective Genetic Algorithm NSGA-II (Non-dominated Sorting Genetic Algorithm II) was used. Optimization result shows that by increasing the volute cross-sectional area from the volute tongue till the circumferential angle 180 deg. provides lower loss. This is due to the improvement achieved for the better distribution of the velocity gradient within the volute. A validated unsteady computational fluid dynamics (CFD) was also employed to investigate the performance difference between optimized volute design and the BASELINE which correlated to the pressure fluctuation and secondary flow behavior inside the cross-sections from 80% to 120% of nominal flow rate. The result shows that the flow distortion in the streamwise direction is stronger with the BASELINE and sensitively affects the operation stability. This is due to the different secondary flow pattern in the cross-sections, hence demonstrating a design direction of desired volute cross-sectional shape for high-performance can be used in a double-suction volute pump.

In most cases of high specific speed mixed flow pump applications, it is necessary to control off-design performance such as shutoff power/head and unstable characteristic as well as design point performance. The authors have been working on multi-objective optimization strategy of mixed flow pump design considering off-deign performance by means of Computational Fluid Dynamics (CFD). In the design optimization process, it was found that the steady CFD analysis using one pitch blade passage adopting periodic boundary condition could be used for relative comparison of the important performance characteristics such as the level of efficiency, the shutoff performance and the stall characteristics of different designs. However, the steady CFD analysis with one pitch blade passage showed that absolute values of head and shaft power were estimated lower than those of the experimental results especially in the partial capacity range. In order to improve the accuracy of CFD results it should be necessary to use full pitch model. In this paper, the evaluation results of three CFD approaches on the capability of the performance prediction of the mixed flow pumps will be shown. The approaches evaluated are steady flow analysis with one pitch blade passage using periodic boundary conditions, full pitch steady flow analysis and unsteady flow analysis. It was found from the evaluation results that the full pitch steady flow analysis showed the same tendency as one pitch analysis and the unsteady CFD provided higher accuracy of the shutoff head. However, the steady analysis should still be useful to reduce the high computational cost and the amount of time. Meanwhile the unsteady analysis clarifies the details of the off-design flow patterns. The effects of the turbulence models and the details of the off-design flow patterns were also discussed in this paper.