Trans. Japan Soc. Aero. Space Sci.Vol. 44, No. 145, pp. 155–163, 2001

Developmental History of Liquid Oxygen Turbopumps for the LE-7 Engine∗

By Kenjiro KAMIJO,1) Hitoshi YAMADA,2) Norio SAKAZUME3) and Shogo WARASHINA4)

1)Tohoku University, Institute of Fluid Science, Sendai, Japan2)National Aerospace Laboratory, Kakuda Research Center, Kakuda, Japan

3)National Space Development Agency of Japan, Tokyo, Japan4)Ishikawajima-Harima Heavy Industries Co., Tokyo, Japan

(Received February 2nd, 2001)

The first stage of the H-2 rocket used a 110-ton thrust liquid oxygen, liquid hydrogen, pump-fed engine, the LE-7. This engine required high-pressure and high-power liquid oxygen and liquid hydrogen turbopumps to achieve thetwo-stage combustion cycle in which the combustion pressure is around 13 MPa. Furthermore, it was very important tooperate both turbopumps at higher rotational speeds to obtain a smaller, lighter-weight engine because the LE-7 had notlow-speed, low-pressure pumps ahead of both the main pumps. The present paper shows the design, test results, andmodifications that had been performed until a flight-type liquid oxygen turbopump for the LE-7 engine was completed.The liquid oxygen turbopump had been developed by the use of three models, that is, research, prototype, and flightmodels.

Key Words: Rocket, Turbopump, Inducer, Cavitation, LE-7

1. Introduction

The H-2 rocket, Japan’s previous expendable launch ve-hicle, which was capable of placing a two-ton payload intoa geostationary orbit, had been successfully operated in sixflights since its first flight in 1994. The seventh flight, how-ever, was unsuccessful because of the failure of the inducerof the liquid hydrogen pump. This failure was thought to becaused by the superposition of some complicated phenom-ena in the inlet portion of the liquid hydrogen pump, whichoccurred mainly because of cavitation and backflow of theinducer.

The first stage of the H-2 rocket used a 110-ton thrust liq-uid oxygen, liquid hydrogen, pump-fed engine, the LE-7. Toobtain high performance, a two-stage combustion cycle wasemployed in the engine. The LE-7 engine required high-pressure and high-power liquid oxygen and liquid hydrogenturbopumps to achieve the two-stage combustion cycle inwhich the combustion pressure is around 13 MPa. Further-more, it was very important to operate both turbopumps athigher rotational speeds to obtain a smaller, lighter-weightengine because the LE-7 engine had no low-speed, low-pressure pumps ahead of both the main pumps. The rota-tional speeds of the liquid oxygen and hydrogen turbopumpswere 18,300 and 42,500 rpm, respectively.

The present paper shows the design, test results, and mod-ifications that had been performed until a flight-type liquidoxygen turbopump for the LE-7 engine was completed. Theliquid oxygen turbopump had been developed by the use ofthree models, that is, research, prototype, and flight models.

c© 2001 The Japan Society for Aeronautical and Space Sciences∗Presented at 36th AIAA/ASME/SAE/ASEE Joint Propulsion Conference& Exhibit, Huntsville, Alabama, July 16–19, 2000.

The research model was fabricated to clarify the major de-sign parameters of the liquid oxygen turbopump related tothe hydrodynamics and mechanical configuration. The pro-totype model was developed to modify the defects that werefound in the research model. The flight type model was pro-duced by minor changes in the prototype model.

2. Design of LE-7 Liquid Oxygen Turbopump

2.1. Mechanical integration1)

The major specifications of the LE-7 liquid oxygen tur-bopump is presented in Table 1. The three types of the tur-bopump are shown in Fig. 1. Since the rotational speed ofthe turbopump was closely related to the weight of the firststage of the H-2 rocket, a parametric investigation was car-ried out to optimize the relationship between the rotationalspeed and inlet flow coefficient of the inducer.1) The liquidoxygen tubopump for the LE-7 has some features in mechan-ical configuration. The simplification of the rotating assem-blies was especially emphasized in the design to avoid rotordynamic problems.

The liquid oxygen turbopump consists of a main pumpand a preburner pump that are driven by a single-stage gasturbine, as shown in Fig. 1. The main pump has a single-stage impeller with an inducer. A large flow rate and highersuction performance required an increased inlet diameter ofthe inducer. Therefore the inducer and the main pump im-peller were arranged as shown in the figures. The guidevanes between the inducer and the main impeller are usefulto support a housing for self-lubricated ball bearings. Withthe connection of the main and preburner pump impellers,an external diffusing passage was selected because the shaftseal pressure of liquid oxygen would be lower than that of

156 Vol. 44, No. 145Trans. Japan Soc. Aero. Space Sci.

Fig. 1. Three models used in the development of the LE-7 LOX turbopump.

Nov. 2001 157K. KAMIJO et al.: Developmental History of Liquid Oxygen Turbopump

Table 1. Major specifications of LE-7 LOX turbopump.

Rotational speeds, rpm 20,000

Main pump

Required NPSH, m 30

Mass flow, kg/s 229.1

Pressure rise, MPa 20.9

Efficiency, % 75

Preburner pump

Mass flow, kg/s 43.8

Pressure rise, MPa 11.4

Efficiency, % 65

Turbine

Power, kW 6,400

Gas inlet pressure, MPa 23.5

Pressure ratio 1.43

Inlet temperature, K 970

Efficiency, % 48.5

Fig. 2. Critical speeds of LOX turbopump.

the internal crossover passage.To minimize the overhang of a turbine rotor, a single-

stage gas turbine is employed at the cost of turbine efficiency,which results in smaller shaft vibrations than those in a two-stage gas turbine. The turbopump could be designed so thatthe nominal rotational speed is less than the second criticalspeed, as shown in Fig. 2, because the second critical speedhas a mode in which the liquid oxygen pump impellers (in-cluding the inducer) whirl.

The axial thrust of the rotor assembly is regulated by a bal-ance piston mechanism as shown in Fig. 3. The pressure ofthe balance piston cavity is controlled by two orifices formedby a back-shroud of the main impeller and a casing. Theturbopump uses a purge of high-pressure, low-temperaturegaseous hydrogen to prevent the turbine’s working gas (hy-drogen rich hot gas) from entering the shaft seal system, asshown in Fig. 4. Self-lubricated ball bearings are cooled byliquid oxygen that passes through filters with fine meshes setin the coolant passages.

Fig. 3. Balance piston characteristics.

2.2. Major component design1)

The major inducer design parameters are presented in Ta-ble 2. The inducer and its guide vanes were designed to usehelical blades. The blade profile of the inducer consists ofa straight line at the entrance and a circular arc. This in-ducer is characterized by a low flow coefficient that requiresa small inlet angle. This angle requires a sharp leading edgeto reduce blockage resulting from cavitation to achieve goodsuction performance. A large swept-back angle was neces-sary to reduce stress near the root of the blades. The inducerwas machined from heat-resistant alloy (Inconel 718).

Both the main and the preburner pumps have three-dimensional impellers designed by only straight lines, usinga ruled surface method. This made it fairly easy to fabricatethe impellers and to analyze flows through the blade pas-sages. Figure 5 shows the main pump impeller.

The turbine blade profile was designed by using the pre-viously reported method. A partial admission nozzle theresearch model employed was changed to a full admissionnozzle in the prototype model because cracks occurred at

Table 2. Design parameters of LE-7 main pump inducer.

Rotational speed N , rpm 20,000

Required NPSH, m 30.0

Suction specific speed S, m, m3/s, s−1 2.10

Cavitation number σ 0.017

Number of blades 3

Inlet flow coefficienta φ1 0.083

Outlet flow coefficienta φ2 0.104

Inducer head coefficienta ψ 0.097

Tip diameter Dt , mm 149.8

Inlet tip blade angle βt1, deg 7.50

Outlet tip blade angle βt2, deg 9.50

aValues for 1.07 times the quantity of nominal flow.

158 Vol. 44, No. 145Trans. Japan Soc. Aero. Space Sci.

Fig. 4. Shaft seal system of LOX turbopump.

Fig. 5. Main pump impeller.

the roots of the turbine blades. The blades were subjected tocyclic loads at both ends of the partial admission nozzle arcs.Furthermore, although the blades and the disk were madefrom a solid material in the research model, the blades wereseperated from the disk in both the prototype and flight mod-els to alleviate the stress resulting from pressure fluctuationsand thermal shock at the start and cutoff of the turbopump.

3. Hydraulic Performance2–4)

The suction performance of the main pump was almostthe same as the predicted one. Figure 6 shows the head co-efficient curve of the inducer of the main pump. Crosses (+)indicate values when the original inducer housing was used,and circles (©) represent values when the modified inducerwas used. The original inducer housing caused the unstablehead coefficient curve, which will be mentioned later.

The overall performance of the main and preburner pumpsis shown in Figs. 7 and 8. Pump efficiency was estimatedby adiabatic efficiency, which was obtained by making useof the temperatures and pressures measured at the inlet and

outlet of both pumps. The main pump of the research modelshowed a slightly higher head and efficiency than those ofthe prototype model. Six small holes were newly fabricatedin the back-shroud of the main pump impeller of the pro-totype model to modify the characteristics of the axial thrustbalance of the rotor assembly, which will be mentioned later.Since the holes increased the internal leakage, the efficiencyand head decreased in the prototype model.

Figure 9 shows the relationship between the turbine ef-ficiency and the isentropic velocity ratio. The turbine effi-ciency was obtained by making use of the outputs of bothpumps. The turbine with a full admission nozzle of the pro-totype model showed a slightly lower efficiency than thatof the turbine with a partial admission nozzle, which mightbe due to the reduction of blade height to 9.4 mm, from15.8 mm.

4. Modifications of the Turbopump

4.1. Regulation of axial thrust balance5, 6)

Much effort was made to establish a balance piston systemin which the back-shroud of the main impeller is used as abalance disk, as shown in Fig. 10. In the initial phase of de-velopment using the research model, the pressure in the bal-ance piston cavity between the two orifices was much higherthan predicted, which also caused a bigger balance pistonforce than predicted. Figure 11 shows the test results withthe relationship between the rotational speed and inlet orificeaxial clearance (without balancing holes A). The inlet-orificeaxial clearance at the design rotational speed was small, anda modification was required to increase the clearance. Thebalancing holes were newly fabricated at the back-shroud ofthe main pump impeller to decrease the pressure in the bal-ance piston cavity as shown in Fig. 10. The addition of bal-ancing holes A made the inlet-orifice axial clearance largeenough, as shown in Fig. 11.

A great amount of time was necessary to find a cause of a

Nov. 2001 159K. KAMIJO et al.: Developmental History of Liquid Oxygen Turbopump

Fig. 6. Suction performance of LE-7 main pump inducer.

Fig. 7. Overall performance of main pump.

Fig. 8. Overall performance of preburner pump.

Fig. 9. Overall performance of turbine.

Fig. 10. Balance piston details.

160 Vol. 44, No. 145Trans. Japan Soc. Aero. Space Sci.

Fig. 11. Inlet-orifice axial clearance.

Fig. 12. Groove geometry.

Fig. 13. Effect of grooves on balance piston characteristics.

Fig. 14. Displacement measurement probe.

higher pressure in the balance piston cavity than expected. Itwas concluded that the disagreement between the measuredand predicted pressure was due to the annular grooves with arow of fastening bolts that were set on the wall of the casing,as shown in Fig. 10. The grooves suppressed the rotating ve-locity of the fluid and made the pressure higher than withoutgrooves. The grooves functioned the same as a swirl breaker.However, it was also confirmed that the balance holes hadanother function to relieve the balance piston cavity pressureduring a stop of the turbopump, which was performed in ashort time and produced excessive pressure in the balancepiston cavity because of the evaporation of liquid oxygen.

Furthermore, the axial clearance of the inlet orifice at thedesign rotational speed was increased by the use of a kindof swirl breaker that consisted of many radial grooves fabri-cated on the casing of the front shroud (Figs. 10 and 12) bywhich the pressure in the balance piston cavity greatly de-creased as a result of the suppression of flow rotation. Fig-ure 13 shows the effect of the grooves on the balance pistoncharacteristics. In that figure, Q0, q0, r2, Tp, u2, and ρ arethe main pump flow rate, balance piston leakage, radius ofmain pump impeller, pump axial thrust, tip velocity of mainpump impeller, and density of pump fluid, respectively. Thegrooves were very effective in increasing the inlet orifice ax-ial clearance.4.2. Supression of rotating cavitation3, 7)

When the original inducer housing was used, it exhibitedthe unstable head coefficient curve presented by crosses inFig. 6. Remarkable head degradation was present near thecavitation number, σ = 0.02–0.04. As shown in Fig. 15,the inducer also produced a supersynchronous shaft vibra-tion and an amplitude jump of synchronous vibration at thesame range of cavitation numbers, which were measured bya displacement measurement probe shown in Fig. 14. In par-ticular, both the largest head degradation and the amplitudejump of synchronous vibration in Fig. 15 occurred simulta-neously, that is, at the same cavitation number, σ = 0.027.From a comparison of the facts mentioned above, a former

Nov. 2001 161K. KAMIJO et al.: Developmental History of Liquid Oxygen Turbopump

Fig. 15. Spectrum analysis of main pump impeller displacements (inducer housing A).

Fig. 16. Details of inducer housing.

report of rotating cavitation, and the experimental investiga-tion of hydrodynamically induced shaft forces with an in-ducer, it was concluded that the shaft vibration was causedby the rotating cavitation that occurred in the inducer of themain pump.

It was conjectured that rotating cavitation might be closelyrelated to the tip leakage flow cavitation of an inducer fromthe visual observations. Some efforts were made to influence

Fig. 17. Main pump impeller displacement in LE-7 engine test.a) original inducer housing, b) modified inducer housing.

the tip leakage flow cavitation of the inducer. Although in-creasing the tip clearance was fairly effective in decreasingthe amplitude of the supersynchronous vibration, it couldnot completely distinguish the vibration. A suction ring,which is usually used to regulate the back flow at the inducerinlet, was also very effective in suppressing the supersyn-chronous shaft vibration. Sometimes the vibration was com-

162 Vol. 44, No. 145Trans. Japan Soc. Aero. Space Sci.

Fig. 18. Fourier analysis of LE-7 LOX turbopump shaft vibrations.

pletely eliminated, but the ring was not applied to the flightmodel turbopump because it required many tests to confirmits durability.

The influence of the inducer upstream housing diameteron the supersynchronous shaft vibration was investigated.Some relationship was found between the inducer upstreamhousing diameter and the amplitude of the supersynchronousshaft vibration. We obtained a very interesting relation ofthe inducer housing dimensions represented by the follow-ing equation, which almost completely extinguished the su-persynchronous vibration, that is, the rotating cavitation.

D1 ≥ D2 + 2C2 = D2 + (D2 − Dt ) (1)

where C2 is the tip clearance and D1, D2, and Dt are de-noted in Fig. 16. The head coefficient curve presented bycircles in Fig. 10, which was obtained by using the modified

inducer upstream housing, did not exhibit the dented partcaused by the rotating cavitation. It was also comfirmed thatthe modified inducer upstream housing was very effectivein suppressing the supersynchronous shaft vibration in theLE-7 engine test, as shown in Fig. 17. This device (inducerhousing C) was applied to the flight model turbopump, sinceit had no durability problems.

5. Other Problems

We experienced a very curious phenomenon in the ini-tial phase of the development of the LE-7 LOX turbopump.Three types of shaft vibrations appeared in the tests of theturbopump alone in almost the same operating conditions.8)

One was a supersynchronous shaft vibration resulting from arotating cavitation that was already described in the previoussection. Figure 18 shows that only a supersynchronous shaftvibration occurred and only a subsynchronous shaft vibra-tion occurred in the almost the same operating conditions.Furthermore, the supersynchronous shaft vibration appearedjust after the subsynchronous shaft vibration had almost dis-appeared concomitant with the decrease of the inducer inletpressure.

The subsynchronous shaft vibration had not appearedwhen the modified inducer upstream housing was employed.Later, an analysis was performed to clarify the causes ofthe subsynchronous shaft vibration. It was concluded thatit was caused by cavitation surge.8) Furthermore, an analyti-cal study of the flow instabilities of turbomachines indicatedthat the rotating cavitation and the cavitation surge occur inalmost the same operating condition.9)

In the initial phase of development, the vanes of a turbinenozzle that were attached to a turbine manifold casing withwelding were separated from the casing, and a large amountof leakage of turbine working fluid from the inlet to the outletof the nozzle occurred, which resulted in the large decreaseof turbine output. This defect was improved by an increaseof the welding area between the tips of the vanes of the tur-bine nozzle and the manifold casing.

Many other minute improvements were performed to in-crease the reliability and durability of the LE-7 liquid oxygenturbopump, such as adding a bypass conduit to increase thecoolant for the turbine-side self-lubricated ball bearings.

6. Concluding Remarks

The liquid oxygen turbopump of the research, prototype,and flight models for the LE-7 engine had been fabricated,tested, and modified from 1986 to 1993. Although the tur-bopumps attained almost the hydraulic performance theywere expected to, regarding mechnical performance someefforts should be made for the flight type turbopump withenough reliability and durability. Some modifications wererequired to achieve the axial thrust balance of the rotor as-sembly. The turbine blades had to be changed from inte-grally machined blades with a disk to a fire-tree blade at-

Nov. 2001 163K. KAMIJO et al.: Developmental History of Liquid Oxygen Turbopump

tachment to reduce the thermal stress at the blade roots dur-ing engine cutoff. With regard to the supersynchronous shaftvibrations, we had to start by clarifying their cause, whichwas found to be a rotating cavitation that occurred in the in-ducer of the main pump. The vibrations were suppressed bya simple modification of the inducer upstream housing.

References

1) Kamijo, K., Hashimoto, R., Shimura, T., Yoshida, M., Okayasu, A.and Warashina, S.: Design of LE-7 LOX Turbopump, Proceedings ofthe 15th International Symposium on Space Technology and Science,1986, pp. 347–355.

2) Kamijo, K., Yoshida, M., Watanabe, R., Hashimoto, Ohta, T. andWarashina, S.: Development Status of LE-7 LOX Turbopump, Pro-ceedings of the 16th International Symposium on Space Technologyand Science, Sapporo, 1988, pp. 281–288.

3) Kamijo, K., Yoshida, M. and Tsujimoto, Y.: Hydraulic and Mechan-ical Performance of LE-7 LOX Pump Inducer, J. Propul. Power, 9(1993), pp. 819–826.

4) Kamijo, K., Yoshida, M. and Nagao, T.: Performance Evaluation ofLE-7 High-Pressure Pumps, J. Propul. Power, 10 (1994), pp. 819–826.

5) Kurokawa, J., Kamijo, K. and Shimura, T.: Axial Thrust Behavior inLOX Pump of Rocket Engine, J. Propul. Power, 10 (1994), pp. 244–250.

6) Shimura, T., Yoshida, M., Hasegawa, S. and Watanabe, M.: AxialTrust Balancing of the LE-7 LOX Turbopump, Trans. Japan Soc. Aero.Space Sci., 38 (1995), pp. 66–76.

7) Tsujimoto, Y., Kamijo, K. and Yoshida, Y.: A Theoretical Analysisof Rotating Cavitation in Inducers, ASME J. Fluid Eng., 115 (1993),pp. 135–141.

8) Watanabe, M., Yamada, H., Yoshida, M., Komatsu, T. and Kamijo,K.: Rotor Vibrations of Turbopump due to Cavitating Flows in In-ducer, Proceedings of (ASME) FEDSM99, 1999 ASME/JSME FluidsEngineering Division Summer Meeting, July 18–23, San Francisco,California, 1999.

9) Tsujimoto, Y., Kamijo, K. and Brennen, C.: Unified Treatment ofFlow Instabilitites of Turbomachines, J. Propul. Power, 17 (2001),pp. 636–643.