Centrifugal Compressor Stage Design Principles Checking

Abstract

Laboratory “Gas dynamics of turbo machines” (LGDTM) has quite effective optimal design computer programs based on theoretic analysis and experimental data. The authors do not share an opinion that 3D impellers are superior in any case. A lot of designed compressors are provided with traditional 2D impellers with cylindrical blades disposed in a radial part of an impeller. The industrial partner tested recently 1:2 scale model of a single stage 32 MWt pipeline compressor. The flow path design is based on the medium specific speed 2D impeller. Good general scheme of the industrial partner, no constrains and profound design optimization have led to maximum efficiency 90% and to excellent performance in a whole. But if a design flow rate coefficient exceeds 0,070 … 0,08 application of 3D impeller is inevitable. Meridian configuration and blade cascade shape of 3D impellers are much more complicated in comparison with 2D impellers. LGDTM has no at its disposal complete information on physical or numerical tests of 3D impeller candidates with different design solutions. Modern trend to apply CFD calculation for investigations to fill the gap seems to be most logical. But the authors’ own experience and published data show that CFD modeling of 3D impeller performance curves is not satisfactory. As a rule calculated performances are shifted to bigger flow rates and work coefficient is 6–9% higher. But the positive moment is that the efficiency at the design flow coefficient is predicted quite accurately. It opens a way to compare stage’s candidates at the design regimes efficiency at the design flow coefficient. The initial design of the stage 3D impeller + vaneless diffuser + return channel with flow rate coefficient 0,105 and loading factor 0,56 is based on general principles of LGDTM: inlet velocity minimization, mean velocity deceleration control, Q-3-D non-viscid velocity diagrams with non-incidence inlet and minimal load at leading edges. CFD calculation has demonstrated necessity to apply a diffuser with tampered initial part, and better shape of the tampered part was defined. The better shape of the crossover was defined by CFD calculations too. The impeller candidates with gas dynamic and geometry principle of blade design, with different degree of flow deceleration, different axial dimension and different exit blade angles were compared. The new 6th version of the optimal design computer programs (Universal modeling was widely presented at the conferences in Japan, Germany, Great Britain, etc.) is tuned on high flow rate stages with 3D impellers. Validation calculations demonstrated good level of performance curves modeling. The program was applied to study series of candidates with different dimensions in meridian plane. As these dimensions influence mean blade load each parameter was studied with different number of blades. Main results are: axial elongation of an impeller does not lead to efficiency grow, optimal leading edge position is at about 25% of meridian distance from an impeller inlet, optimal inlet diameter is 8,5% less that the diameter corresponding to minimal peripheral inlet velocity. The last conclusion is of particular interest and needs additional proof. The comparison of 94 impellers candidates has led to the stage efficiency increase on about 1.5%. The results have verified general principles of design applied in the laboratory “Gas dynamics of turbo machines” and pointed out on some improvements of design principles.