by Christopher E. Brennen

The third, and perhaps final, chapter in the Caltech Pump Lab saga was motivated by problems that arose in developing the high-speed turbopumps for liquid-propelled rocket engines. Early liquid rockets had pressurized fuel and oxidizer tanks that fed these fuels to the combustion chamber but the limitations of this model were soon realized and it became evident that high speed pumps would be necessary to achieve higher power levels. These had to be driven by gas turbines fed by gases from small combustion units known as pre-burners. The units consisting of the pump integrated with the turbine were known as turbopumps and the development of turbopumps for both the fuel and the oxidizer became a critical part of the design of liquid propelled rocket engines. Early in their development it became evident that in order to minimize the mass of these components it would be necessary to run them at high speed and therefore they would cavitate extensively. Consequently a first pump stage designed to cope with cavitation became a neccessary part of the pump design and was known as a cavitating inducer. These inducers cavitated extensively and became the subject of considerable design and performance efforts. Allan Acosta, Professor of Mechanical Engineering at Caltech became involved in this research effort through his contacts with the US Navy (whose interest was in cavitating propellers, a related issue) and Rocketdyne Division of North American Rockwell. During the late 1950s and early 1960s he built an inducer test facility in the Pump Lab in the basement of Guggenheim that produced seminal understanding of the performance and design requirements for cavitating inducers, knowledge which both Aerojet General and Rocketdyne incorporated in their turbopump designs for rockets such as the Titan and the Saturn. That efffort ended when the Guggenheim basement was renovated and that version of the Pump Lab was disbanded.

However, the Caltech involvement with pumps and with the development of liquid-propelled rocket engines did not end there. As the development of liquid-propelled rocket engines proceeded in the 1950s and 60s in both France and the US it became evident that there existed some serious instabilities of the propulsion system that could potentially cause catastrophic failure of the rocket. Perhaps the most critical one of thes instabilities was named the ``Pogo instability'' after the child's toy of the same name. It worked as follows. If the vehicle began to oscillate in a longitudinal structural mode, this would cause pressure oscillations in both the fuel and oxidizer tanks. Thus the engines would experience fluctuating inlet conditions and would therefore produce a fluctuating thrust that might feed back into and amplify the longitudinal structural vibrations. A few examples illustrate the seriousness of this problem. The French Diamente B vehicle suffered several catastrophic failures when the Pogo instability amplitudes reached 10s of g. In 1962, a Titan II rocket was destroyed in flight because 10g Pogo oscillations led to premature shutdown of the first stage engines. With the advent of manned flight, the issue became more critical since it was essential to avoid frequencies and g levels that would disable the pilots. However, the second stage of the Saturn V used in the Apollo missions also suffered from Pogo and on the Apollo XIII one of 5 engines shutdown prematurely due to 33g oscillations at the engine (much less in the manned capsule). Despite this the mission proceeded successfully.

Analyses of this Pogo problem by Caltech Alum Sheldon Rubin in the early 1960 eventually lead to an understanding of the important role that the oscillations of the cavitation volume in the turbopumps could play in the instability. Yet it was clear that there was very limited understanding and almost no reliable data on how the cavitation in a turbopump would respond to fluctuating inlet conditions. At this time the Space Shuttle Main Engine was being developed and the issue of how to prevent Pogo oscillations in that vehicle naturally came to the fore. Thus it was that Loren Gross, one of the members of the NASA Huntsville team with oversight responsibility for the SSME turbopumps paid a visit to Professor Allan Acosta about 1970. Gross's account of the SSME problem renewed Acosta's interest in cavitating inducers and so began a long a valuable connection that led to a much improved understanding of the dynamic behavior of cavitating and the dynamic response of cavitating turbopumps and therfore to informed analyses of the Pogo instability (as well as similar instabilities in many other pumping systems).

Thus a new Pump Lab was born in the basement of the Thomas Laboratory. Acosta and Chris Brennen constructed a unique facility (the DPTF or Dynamic Pump Test Facility) for the measurement of the complete dynamic transfer function for a cavitating pump. The information on those transfer functions (the connection between the fluctuating pressures and flow rates at inlet and discharge over a range of frequencies and cavitation numbers) was used to determine appropriate ameliorative measures in the Space Shuttle and, at least thus far, this has not been a problem with that vehicle. When the Brennen/Acosta facility finally became redundant about 2002, it was disassembled and transported back to Huntsville for reconstruction there; a number of replica facilities have now been constructed elsewhere in the US, Japan and Italy by former students and visitors to the Thomas Pump Lab.

But there is a final chapter to relate for another critical issue with the development of the SSME turbopumps occurred about 1980. During testing of the high pressure hydrogen fuel turbopump it became evident that the critical whirl speeds of that machine were significantly lower than had been anticipated during the design phase and that meant a serious limitation to the speed and power level of the engine. Now whirl speed calculations of that era took little account of the fluid-induced rotordynamic forces in the machine that might effect the critical speeds and it was rightly judged that there was a serious missing component in those calculations that had to explored and rectified. One possible source of error was in the fluid-induced rotordynamic response of the seals and Dara Childs at Texas A and M took responsibility for that part of the problem. Chris Brennen and Allan Acosta took responsibility for the investigation of the contribution from the main flow through the turbopump and constructed another facility (called the RFTF, the Rotor Force Test Facility) to measure all the fluid-induced rotordynamic forces due to the flow through a pump impeller with and without cavitation. The data and understanding collected during that research effort constituted an important new feature for the design of high-speed turbomachines.

The Thomas era of the Caltech Pump Lab ended in 2003 when both the DPTF and the RFTF were transported back to the NASA George Marshall Space Flight Center in Huntsville, Alabama.

Last updated 3/9/07.

Christopher E. Brennen