The Delta 4 rocket's cryogenic upper stage engine persevered through a fuel leak and the resulting low-thrust condition by autonomously engaging techniques that kept the mission on track to successfully launch a Global Positioning System satellite in October.

Animation of the upper stage firing. Credit: United Launch Alliance

The dramatic ascent began at 8:10 a.m. EDT Oct. 4 from Cape Canaveral's Complex 37, roaring into the morning sky on the combined power of the RS-68 main engine and twin solid motors.

It was a flight being conducted by United Launch Alliance for the U.S. Air Force to deliver a next-generation GPS navigation satellite into orbit where it would replace an aging, 19-year-old craft.

But four-and-a-half minutes into the launch, after the first stage had shut down and separated, the trouble began as the RL10B-2 engine on the upper stage extended its nozzle and fired to life.

When the powerplant was igniting and reached its peak chamber pressure, a leak started above the narrow throat portion of the thrust chamber, setting off a series of ramifications that would endure over the next three hours as the vehicle made its climb to the GPS constellation.

The mission sequence planned three burns of the upper stage, initially reaching a low-altitude parking orbit, then a highly elliptical transfer orbit and eventually achieving a circular orbit in line with the GPS network 11,000 nautical miles up.

Throughout those burns, however, the engine was feeling the effects of the leak by producing less thrust than expected, firing longer than planned to compensate and having to adjust its trajectory.

"With the fuel leak, the thrust was lower than nominal, and the Delta 4 closed loop guidance system measured this in real time and revised the trajectory that was being flown and also the burn durations to achieve the required orbital conditions and other requirements," said Jim Sponnick, ULA vice president of mission operations.

"Also, the fuel leak in the thrust chamber resulted in a slight offset to thrust direction from the engine. The launch vehicle control system adapted to this by adjusting the engine gimbal angle to compensate for the slight offset in the thrust."

Despite all of that, the rocket's resilience allowed the launch to achieve success in the face of adversity, hauling the 3,400-pound GPS satellite exactly where the cargo wanted to go.

"We were later informed by our GPS customer that this was the most accurate placement of the three GPS 2F spacecraft that have been launched, which enabled a short on-orbit checkout period following the launch," Sponnick said.

Using the margins in the rocket's fuel supply and calling upon the flight software to handle the off-nominal situation, the rocket kept clawing onward on each of its three burns.

"The three burn durations ranged from 20 to 36 seconds longer than nominal predictions. Various parameters are targeted for each burn, and the flight results generally did meet our expectations. The final orbit was quite precise," Sponnick said.

The first burn was supposed to last nearly 8 minutes to put the vehicle into a low-altitude parking orbit. The rocket coasted over the central Atlantic for about 9 minutes before restarting the engine to run for a scheduled three-minute firing to inject itself into an elliptical orbit with a high point near the GPS constellation's altitude 11,000 nautical miles, while leaving the low point around 130 nautical miles.

Animation of the upper stage firing. Credit: United Launch Alliance

On the ground, mission managers realized they had a problem on their hands but were not quite sure how the ascent was going to turn out. A worst-case scenario was the upper stage running out of fuel before getting to the proper orbit, casting fears of a "bad day."

The rocket was in the midst of a quiet, three-hour coast as the rocket motor traveled away from Earth toward the high-mark of the orbit for one final firing.

"While we knew in real-time about the lower than expected engine performance, we were not able in the limited time to perform detailed analyses that would be needed to accurately quantify the propellant leak. As a result, we had rough analyses ranging from predictions of a nominal final orbit to some conservative worst-case analyses showing that there might not be adequate performance to achieve the required final orbit," Sponnick recalls.

The third burn ignition time arrived and the RL10 again restarted, giving all of the propulsion needed to fully circularize the orbit by raising the low point up to the GPS network's altitude.

"The thrust in the first burn was approximately 5 percent below nominal, and it stepped down several percent more in each of the second and third burns. During each of the three burns, the thrust levels were constant," said Sponnick.

Three-and-a-half hours after leaving the launch pad, Delta released the satellite cargo into an approximate 11,047-nautical-mile perch tilted 55 degrees to the equator, marking a full mission success.

"It was a relief to the team that there was plenty of propellant margin and the final orbit was accurately achieved," Sponnick says.

And, in fact, there was "pretty substantial margins" still left in the tank when the mission was completed. Analysis now shows more than 1,000 pounds of propellant remaining, which would have enabled approximately 30 seconds of additional burn time, if necessary, Sponnick says.

Fed with supercold liquid hydrogen and liquid oxygen, the RL10B-2 is the latest in a long line of upper stage engines dating back a half-century. The original version of the RL10 debuted successfully on an Atlas rocket in 1963 and has been part of Centaur for more than 200 space missions.

The RL10 has dispatched robotic expeditions to every planet in our solar system, plus multiple missions to the moon and countless military spacecraft and commercial communications satellites in orbits around Earth.

This latest RL10 variant was introduced in 1998 as part of Boeing's Delta 3 program, which served as a stepping-stone to the Delta 4 rocket and development of its cryogenic upper stage.

The engine has been fired in space 23 times to date.

Its specs include a nominal thrust of 24,750 pounds, mass of 664 pounds, an overall length of 13.6 feet, including 7 feet just for the nozzle extension and a specific impulse of 465.5 seconds.

An upper stage of the Delta 4 rocket. Credit: NASA

Coupling the Delta 4 with the relatively light-weight GPS 2F satellite provided a bit greater margin for the launch than other rocket and payload combinations.

"There are missions that would not have reached their prescribed orbit with the kind of performance degradation that occurred in the last launch," Sponnick says. "That illustrates why we must thoroughly investigate and implement corrective actions for an anomaly such as this."

Investigators used extensive analysis and the reconstruction of flight data to find the leak location and when it began. But the ongoing inquiry continues to work through the credible "candidate causes" to determine why the leak happened.

The team had 2,000 measurements from the GPS launch that were analyzed in great detail and technicians performed tests on production engines to aid the investigation and crossover assessments. More than 500 "candidate causes" have been thoroughly assessed in the process of zeroing in on a small number of credible physical causes.

Pratt & Whitney Rocketdyne produces the RL10B-2 engine and the RL10A-4 powerplant used by the Centaur upper stages on Atlas 5 rockets. Although the two engines have their differences, they do share commonality through the evolution of the venerable RL10.

The list of "candidate causes" of the leak includes some pertaining only to the Delta 4 version and others that have "crossover" implications that could touch Atlas as well. Those potential causes common to both vehicles have been put to the test against the Atlas engine that will power Tuesday's launch of the X-37B spaceplane, known as Orbital Test Vehicle Flight 3, and officials Friday formally cleared that RL10A-4 for flight.

"For each cause that has potential crossover to the RL10A-4 engine, comprehensive flight clearance assessments have been performed for the OTV 3 mission," said Sponnick. "Where applicable, specific mitigating actions have been implemented -- such as detailed visual and borescope inspections of the engine and launch vehicle systems."

The Atlas 5 rocket will be wheeled to the launch pad Monday for a targeted blastoff Tuesday at 1:03 p.m. EST, weather permitting. See our Mission Status Center.

It will get the Atlas family back in action, which paused its manifest in light of the RL10 situation and delayed the X-37B mission from the targeted late October launch.

The Delta 4 rocket had planned mid-January for its next mission will carry the Wideband Global SATCOM 5 military communications satellite to orbit. That vehicle's engine has undergone the same visual inspections that the X-37B launcher did, and some additional work on the Delta is being evaluated before it is cleared for flight. A new launch date is pending.

Atlas plans to launch NASA's Tracking and Data Relay Satellite K on Jan. 29 from Cape Canaveral and return to its original schedule with the Feb. 11 flight from California's Vandenberg Air Force Base carrying the next Landsat remote-sensing spacecraft.