Launch failures: fill ’er up?

A Proton launch in 2010 failed not because it ran out of propellant but instead because it had toomuch on board. (credit: Roscosmos)

One of the most common causes of airplane accidents is a pilot sitting there and letting the thing run out of gas. This type of mishap is much less common with space launches, but early propulsion system shutdowns due to the vehicle running out of propellant have occurred in some noteworthy cases.

The majority of liquid propellant space boosters ever launched have lacked a system with even as little sophistication as a bewildered pilot staring at a dropping fuel gauge. The engines were tested, the performance noted, and the required amounts of fuel and oxidizer calculated using simple formulas. For vehicles using liquid oxygen (LOX) as the oxidizer, that tank was topped off: a necessity since it kept boiling off until mere seconds before liftoff, when the vent valve was closed. The fuel was loaded based on the calculations, with a bit extra added to provide some margin. Thor, Titan, and Delta all used this approach, as did most foreign vehicles.

The Atlas was different, employing a Propellant Utilization System that measured the amount of fuel and oxidizer in the tanks and adjusted the engine thrust to maximize performance. However, this only occurred during the sustainer phase of the flight: the big booster engines just ran at whatever thrust levels they had been set to for about the first two minutes of flight. The system kicked in once the vehicle was running on the central sustainer engine. It was this system that enabled Atlas 19F to recover from a serious performance loss on the NOAA-B mission on May 29, 1980.

The challenge of calculating the correct propellant load was well illustrated by the failure of the Defense Meteorological Satellite Program Thor LV-2F F34 mission launched from Vandenberg AFB on February 19, 1976. At the time, the DMSP launches on Thor boosters used an excessively simple process for calculating fuel loads. The performance test data for the booster’s first stage engine was used by the launch crew to calculate the required fuel load and a counter was used to measure the amount of fuel loaded in the vehicle. That was it. The required amount of fuel was loaded into the booster, the countdown proceeded, and liftoff occurred. But the payload did not make a stable orbit, re-entering by the end of the first orbit.

The subsequent investigation revealed that the performance test data provided for the engine was wrong. The engine required more fuel to achieve the required performance than the data indicated. The actual situation was as if you walked onto a car lot filled with new automobiles, read the data stickers on the windows, and chose a model that got 39 miles per gallon (mpg) rather than the 34 mpg every other car of that exact make, model, and options listed, and then did not wonder why that car was so much more efficient. There was no one making that kind of comparison. Studies were done and a much more rigorous performance analysis was conducted for the remaining Thor launches.

Another failure that occurred at Vandenberg AFB was the Delta 3914 Dynamics Explorer mission of August 3, 1981. Normal procedures required the Delta second stage to be loaded with propellant during the countdown. For that mission, a new feature was added to the propellant loading system: a “pinwheel” that rotated in the feed line, indicating flow just like the pinwheels that used to be displayed at some gas station pumps. Unfortunately, the new pinwheel jammed and a leak also developed, leading the launch crew to conclude the second stage was full of fuel. The second stage ran out of fuel about 16 seconds early, placing the payload in an orbit about 100 miles (160 kilometers) low. As it turned out, there had been a disagreement among researchers over the orbit to be used, so those who had argued for a lower orbit were happy, but the rest were not.

It was a big day for the Indian space program when its first Geosynchronous Launch Vehicle, GSLV, launched the GSAT 1 mission on April 18, 2001. There was celebration at the achievement, but it was short lived. The third stage used a new Russian-built engine that had not flown before and was deficient in performance. The spacecraft attained an orbit that was about 0.5% low on velocity, resulting in the spacecraft being unable to attain its required orbital slot. While operating normally, the spacecraft was drifting below and across other orbital positions, interfering with one communications satellite after another. This was unacceptable and the spacecraft was turned off after only a few days.

A new version of the venerable Proton booster lifted off from Baikonur Cosmodrome on December 6, 2010 for a GLONASS navigation system mission. The vehicle had the new DM-03 higher performance upper stage. The payloads never made it to orbit, impacting in the Pacific Ocean. It turned out that the situation was the opposite of the Delta Dynamics Explorer mission. The new DM-03 upper stage had considerably more propellant tankage than the earlier versions and the loading procedure did not reflect that fact. Although the mission did not call for the added propellant, it was loaded anyway, a total of about 2,000 kilograms more than was required. Instead of too little propellant as in the Thor F34 and Delta Dynamics Explorer failures, the Proton had too much.

Why was too much propellant a problem? In the case of the Thor F34 failure it was not just a case of there not being enough fuel on board. For the DMSP missions the Thor and the upper stages used were almost too limited for the spacecraft, the mass of which kept getting larger with every mission. One approach used to address this performance concern was to replace the RP-1 fuel used with RJ-1 fuel. Designed for ramjet use, RJ-1 was denser than RP-1, and that meant that more pounds of fuel could be packed into the limited volume of the Thor fuel tank, and that in turn translated into more energy.

For the Thor F34 mission the apparently superior performance of that Thor’s engine was noted years in advance and thus it was specifically selected for the heaviest mission of the spacecraft series. The simple fact was that not only did that particular engine not have that kind of performance, no engine of that general design ever built did, either. It was impossible to put enough fuel in the Thor tank for the mission to succeed because the increased weight of fuel would have degraded the performance so much that it would have failed to attain the proper orbit anyway.

This same problem applied to the Proton DM-03. It had plenty of propellant in the upper stage but for the trajectory employed it was too heavy as a result. In the design of the Delta IV and Atlas V vehicles, development and production costs were the major drivers and one thing that was clear was that engines cost a lot more than did propellant tankage. Previous vehicles using RL10-powered upper stages had used at least two of the engines, but with proper trajectory design just one engine could be used. The trajectory would be lofted almost straight up during the first stage burn, thereby getting away from both aerodynamic drag and the gravity losses associated with a lower and more efficient trajectory. Once up high, the RL10 upper stage could burn for a long time, gaining velocity much more slowly but also saving a lot of money in terms of expensive hardware. This approach produced some range safety concerns but since US launch ranges fly over the ocean it was not insurmountable.

Presumably the Proton DM-03 mission of December 6, 2010, could have flown a similar lofted trajectory and placed the upper stage at an altitude where it could use its extra propellant, but no one thought of that because the stage was not supposed to be loaded that full of propellant anyway.

So, it is not just a problem of being sure to fill up the tank before you hit the road, but rather one of making sure you have the right amount for the mission. And be sure to look at more than one sticker on those car windows before you pick one out.