logo SBA


Digital archive of theses discussed at the University of Pisa


Thesis etd-11242014-192002

Thesis type
Tesi di dottorato di ricerca
Thesis title
Academic discipline
Course of study
tutor Prof. D'Agostino, Luca
  • Cavitation
  • dynamic matrix
  • inducer
  • pogo
  • rotordynamics
  • turbomachines
Graduation session start date
Release date
Despite the turbomachines are used in space propulsion since the beginning, they still represent crucial components for liquid-propelled rockets. The complexity of these devices requires to pay particular attention to their design behavior, especially in presence of cavitation. Indeed turbomachines are usually asked to work with a two-phase flow which can affect not only the pumping performance, but can also generate peculiar flow instabilities closely connected with the cavitation and whose amplitude depend on the cavitation level.
Cavitation is at the basis of one of the most important technical problems faced in rockets powered by turbomachines: the arising of the POGO phenomenon. This phenomenon has been known since the dawn of the space missions and it consists of a resonance where the dynamics of the longitudinal vibrations of the rocket structure can couple with the thrust oscillations eventually leading to mission failure. The POGO oscillations are very related to the presence of cavitation which can alter the dynamic response of the turbomachines to pressure and flow rate oscillations generated in the ducts by flow/combustion instabilities or other types of instabilities.
In this thesis the changes brought to the test facility at Alta S.p.A. in order to adapt the test rig to the experimentation of the POGO phenomenon will be shown. The innovative system for the generation of the oscillations is based on the vertical vibration of the tank so that pressure oscillations are produced and, thanks to the impedance of the discharge and suction lines, also flow rate oscillations are consequently generated. These oscillations are assessed by means of piezoelectric pressure transducers placed along the lines whose readings have been used for evaluating the dynamic transfer matrix for different pumps.
In particular three different pumps have been tested: two inducers and a centrifugal pump. Both the inducers have been derived from the DAPAMITO inducers family whose design has been drawn by means of the model developed by Professor d’Agostino. The new inducers have been modified in the leading edge shape, whose tip shape has been rounded in order to study its effect on the performance and on flow instabilities, which are notoriously affected by the shape of the leading edge.
The experimental study of these machines outlined weak effects of the leading edge shape on the non-cavitating performance, but greater effect has been noticed on the suction performance which shows an early head degradation. The analysis of the flow instabilities has also shown the effect of the shape of this important part which can affect the presence of some forms of instabilities. Indeed a comparison with the previous inducers and a later one with sharp edges confirmed the effect of this part on the flow instabilities, some of which, like the high-order surge, can disappear with the rounding of the leading edge. Anyway a deeper analysis of the cavitation effects should be carried out given the relatively high values of cavitation numbers tested. This should be performed in order to verify if some phenomena (especially the some rotating instabilities) could arise at lower cavitation numbers thus showing a delaying effect of the rounded leading edge on the flow instabilities.
The study of the flow instabilities on the centrifugal pumps outlined instead the need in the next future to make some changes to the test rig so that the pressure taps can be placed nearer to the pump. Indeed in the actual configuration, because of the distance of the pressure taps, some possible rotating instabilities have been detected as axial.
For the characterization of the dynamic response of the considered pumps the test facility has been modified on the basis of a model previously developed. The model outlined the need to have two different configurations which could give the possibility to perform distinct tests and so to get the dynamic matrix of the pumps. A first test campaign has been performed and it outlined many different problematics which demonstrated to be of crucial importance for the successof the experiment. Among the different problems there have been detected: excessive vibrations of the suction and discharge lines which have been almost completely eliminated by using reinforced concrete plynths to block the two pipes beside the tank; too many air pockets in the water have been detected and the number of the dearinting cycles have been so increased; some differences in the sensitivity of the pressure transducers have been found and so a specific calibration has been performed in order to find a way to properly correct the sensitivity at varying the frequency and the pressure amplitude in sucha a way that all the transducer would have given the same result for the same input.
A second experimental campaign for the dynamic matrix characterization has been performed only on two test items: the three bladed inducer and the centrifugal pump. This experimental campaign, anyway, hasn’t brought the expected results. The vibrating table proved to be able to generate significant pressure oscillations, but the trend of the dynamic matrix elements showed to be very different from the one foreseen by the model or by the few previous experimental results reported in literature. Some models have been also developed and presented in this thesis in order to understand the results obtained. Their capabilities have been compared by fitting the experimentala data available in literature. Each proposed model shows a different complexity in describing the physics involved in turbomachines, but the model which best fits the experimental data has been the one already used in the design of the experiment. So this model has been used in order to better characterize the dynamics of the whole circuit on the basis of the experimental data recorded during the second experimental campaign and it gives the possibility to get the component of the dynamic matrix by fitting the data related to the pressure oscillations at the pump inlet. This model of the circuit has also allowed to assess the very high condition number of the problem and it has highlighted how much some errors on the measurements can influence the assessment of the dynamic matrix. In particular it has been noticed that an error on the evaluation of the amplitude of the pressure oscillations had minor effects than the errors on the determination of their phases which resulted to totally change the trend of the dynamic matrix components at vearying the frequency of the imposed oscillation. The same model has been also used to identify some configurations which can give different geometrically independent configurations by only changing the position or the value of a compliance, which is physically represented by an hydraulic accumulator. In this way different experiments can be performed for the same imposed oscillating frequency and it is so allowed the possibility to reduce the effect of measurement errors on the transfer matrix assessment.
The second part of the thesis is dedicated to the study of another very important problem: the rotordynamic forces acting on turbomachines, especially in the presence of cavitation. The experimental activity has concerned an inducer, similar to those previously tested, a centrifugal pump and their combination. In this thesis, anyway, it is reported only the work performed on the inducer. The effects of the flow coefficient, the intensity of the cavitation and of the temperature on the forces is shown. In the tests it came out that the inducer is weakly affected by the intensity of the cavitation until this becomes such intense to lead the pump head to drop. In this case the force acting on the pump can vary significantly in intensity and also in behavior. The effect of the temperature has been instead almost negligible for all the tested conditions whereas the incidence angle showed some interesting effects on the disposition and the values of the maxima and minima of the rotordynamic force.