Figure 1-5. What Change in What Specifications?
Jupiter Propellant Sloshing Solution. "Beer Cans"
We lost the first Jupiter missile due to an engine plume recirculation problem burning the actuator control wires. The second Jupiter launch was lost due to propellant sloshing dynamics coupling with the control system. [Ryan, September 1996; Abramson, SP-106, 1967; Abramson, SP-8031, 1969].The control system was saturated and vehicle control was lost at max q. The first problem was fixed with the installation of a heat shield where the gas generator exhaust was dumped overboard; however the second problem required more work and engineering creativity. No analytical models existed for characterizing the dynamics of liquids in a tank, and this type of experimentation was an emerging technology. The problem became the instigator for a long term analytical and experimental technology development.
Helmut Bauer at NASA and Norm Abramson at Southwest Research Institute were the leaders in this effort. [Bauer, 1964] However, this effort was downstream and we needed a quick solution so that Jupiter 3 could be launched on time. We took the Jupiter LOX tank and put it on an empty railroad car and filled it with water. We then bumped the railroad car against the spur railroad stop, exciting the liquid dynamics. A movie camera recorded the motion and we were able to derive an equivalent slosh mass and frequency to be used in a control feedback simulation. The question then was: ―What is a fix for the problem?‖ so that we could launch the next vehicle. Someone said that when they were back on the farm and had to haul water in steel drums in a wagon, that they floated pieces of lumber on the surface to keep it from sloshing. Well, lumber would not work in a missile, but floating something would. The original design was a perforated cylinder truncated with cones and had a
commode float inside to make it float. The entire surface was filled with these devices called
Bill Eddy
beer cans. We then put these in the tank on the railroad car and demonstrated that they would indeed suppress the sloshing. They were eventually flown on Jupiter. See Figure 1-6.
Later, through analysis and sub-scale testing, we developed the baffle approach making the perforated baffles a part of the ring stiffeners. See Figure 1-7. This approach was used on Saturn and saved weight by having the baffles also perform part of the stiffening required to prevent tank buckling etc. Shuttle used baffles attached to an inter-frame instead of the ring frames due to manufacturing and operational requirements.
Figure 1-6. “Beer Cans” Flown on Early Jupiters
Figure 1-7. Later Jupiter Configuration with Slosh Baffles
Synthetic Wind Profile
Early in Saturn development, we were challenged with the problem of representing the atmospheric wind characteristics in a manner that would allow 6-degree of freedom control and vehicle dynamic simulations to provide time consistent 3-sigma response data for
structural design. Two creative ideas were required to meet this goal. The first involved a way of taking each single parameter 3 sigma run of the 6-DOF response run and comparing it to a nominal run, extracting the differences for each response parameter. These deltas were then root-sum-squared (RSS‘d) and added to the nominal, producing the 3-sigma design values.
The problem was that this discrete value was needed in combination with all other parameters in the time consistent manner in order to have a balanced load set. Jud Lovingood came up with a way of taking the 3-sigma deltas and ratioing them with the nominal to generate a ratio for the input parameters, which then provided a 3-sigma time consistent response. [Lovingood, J.A. 1964] In addition we needed a way of having the wind characteristics modeled in a time consistent manner for a forcing function for the 6-DOF simulation. Helmut Horn, Jim Socggins, Bill Vaughn and Robert Ryan came up with the synthetic profile based on a 95% wind speed and RSS‘d 99% wind shear and square waved wind gust. See Figure 1-8. This was used very successfully in the early design of Saturn.
[Geissler, E.D. et.al.1970] The need arose to have a more realistic representation of the wind shear and gust, so a monthly set of detailed measurements were measured over a few years timeframe, that had the wind gust and shears correct to 50 meter lengths. Using these
detailed Jimsphere wind profiles a Monte Carlo approach was used to verify that
accelerometer feedback load relief was not effective. As a result Saturn V flew without load relief. (See Lesson 6) [Ryan,R.S. January 20-23,1969; Geissler, E.D. et.al. 1970] Since then a vector wind model and the Global Reference Atmosphere Model (GRAM) wind model have been developed for use in Shuttle and new programs.
Figure 1-8. Synthetic and Measured Wind Profile Approach
The need for and the application of this approach is summarized as:
• Control and Loads response analysis requires time-consistent data.
• Question: How do I generate this time-consistent data from a root-sum-squared (RSS) peak value from perturbated individual response runs?
• The A-Factor approach ratios the individual perturbated peak values to the RSS‘d value, to obtain parameter scaling factors, which will produce a time consistent run with the peak value matching the RSS‘d. This produces a time-consistent data set for all response variables at a 1-sigma level
SSME LOX Pump Silicon Nitride Bearings
During Space Shuttle operational flights, liquid oxygen (LOX) pump bearings were a major problem, as were other elements such as turbine blades and welds. Bearings would wear out very quickly and along with other problems led to a requirement to refurbish the pumps after every one or two flights. Alternate turbopumps were proposed as a solution to these problems. The development of the alternate LOX turbopump was having major issues with the pump end bearings in that they would overheat and wear out in the first 50 seconds of run time. This problem was threatening continuation of the program, and needed a quick solution. A team was formed to find a solution to the problem. They tried everything with nothing working. Prior to this team‘s formation a MSFC engineer Dr. Robert Thom, was thinking out of the box and came up with use of Silicon Nitride (SiNi) ball bearings and had a technology program in place to test the bearings in a bearing tester. The team was near the deadline of program cancelation unless a solution was found, when we decided to try the new SiNi balls. [Gibson, Jannaf-1354] Silicon Nitride is a ceramic material. A manager of Pratt &
Whitney was reluctant to try them and had said no one would put glass balls in his pump. A Pratt engineer took him into the shop and put a new ball bearing in a nylon sack and had him hit it with a large sledge hammer. The anvil and the hammer were dented, but the ball had no fracture under microscope inspection. As a result we started testing the balls in a pump and surprisingly there was no wear of the balls or the race. The results have been that pumps can fly 20 times with no bearing wear. See Figure 1-9. Currently, all space shuttle turbopumps have these SiNi bearings.
Figure 1-9. Alternate SSME LOX Pump with Silicon Nitride Ball Bearings
Tethered Satellite Skip Rope Damper
Tethered satellites have many technical features that are desirable. The can be used for power sources, orbit changes etc. However, implementing a Tethered Satellite had major design problems. As the satellite was deployed on a tether, the tether would set up dynamic oscillations (skip rope), which in the end could destroy the system. A team was formed to solve the problem through analysis and test. It was clear that some means of damping the oscillations was required. The approach was to install a damping system in the tether output ferrule using negator springs. As the tether dynamically moved, the dampers on the end of the cable attached to the eye through which the tether was deployed would move with the dynamic motion, thus damping the skip rope. After much testing the system was verified and flew very successfully on the first tether flight. See Figures 1-10 through 1-12 for details.
Figure 1-10. Tether Deployment Scenario
Figure 1-11. Tether Skip Rope Characteristics
Figure 1-11. Tether Skip rope Characteristics