Part V Chapter 12:
“We need affordable space travel to inspire our youth, to let them know that they can experience their dreams, can set significant goals and be in a position to lead all of us to future progress in exploration, discovery and fun.”
-Burt Rutan
A little farther out of the public eye than the flashy NASA missions, the military's reusable launch vehicle test program proceeded steadily. The Prometheus launch system was making great progress. Though the team had planned to begin booster tests by 1988, they were delayed by the development of the RL-51. This was not unexpected, as the RL-51 was a supremely complex and impressive engine. The requirements imposed upon the engine by the reusable vehicle programs were daunting. It would need to be the most powerful hydrogen fuelled engine ever built, producing more thrust than the J-2S, with a specific impulse close to that of the RL-10. It would need to be reusable, with minimal refurbishment between flights, and certified for over 100 relights. To support the landing system of the Prometheus Booster, would need to be restartable in the air, and capable of deeply throttling. Despite these incredible requirements, Pratt and Whitney managed to deliver the engine with minimal delays. By early 1989, the first prototype Prometheus booster had been fully constructed, with the engines being one of the last things integrated. It measured 6.6 meters in diameter, the same as a Saturn II first stage, and over 30 meters long, and massed in at 29 metric tons. It given the serial number PB-01T. Though Grumman had built and flown the X-29 all alone, for Prometheus they partnered with Boeing. A plant to assemble the boosters was built in Washington State, just outside of the city of Seattle. This was seen as controversial, as the expensive plant was largely funded by taxpayer money, and, if things went well, would only assemble half a dozen or so boosters. Critics argued that it would be cheaper to use existing facilities, like the Marshall Space Flight Center or the Michoud Assembly Facility. However, the military wanted the vehicle to be built on the west coast, so as to be closer to the intended launch site, and so, like with many military projects, they received the necessary funds. By July, the assembled vehicle was shipped from the factory in Washington, down the Pacific coast, to Vandenberg Air Force Base for the first static fire tests. The first static fire occurred on August 11, testing the structural integrity of the booster, and the capabilities of the engines, including their ability to relight. Several more static fire tests occurred, and in between each one, the vehicle was carefully inspected, to see how well the engines had survived, and how much refurbishment was necessary. The engineers at Pratt and Whitney had done their job, and the RL-51s proved themselves to be robust and reusable.
Finally, in November, the booster was readied for the first “hop” tests. This part of the program would be similar to the flights performed during the X-29 program. In order to prevent damage to the launch pad, these initial hops would be performed at the specially designed landing pad. A bonus benefit of performing the tests here was that the booster recovery infrastructure could also be fully tested, and booster turnaround could be optimised. The huge booster, much bigger than the X-29, was incredibly daunting to ground engineers as it towered over them, like a skyscraper. It had intimidated those designing it just as much. The titanic stage was much bigger than the X-29, and thus would be slower to turn or react in flight. This meant that it needed to much smarter, with the flight software being totally automated, planning out everything in advance, and accounting for many factors, far in advance. The actual landing was a bit more complex than that of the X-29. The booster would need to burn once to return to the launch site, and then again to land. The booster would need to be able to accurately hit the landing pad. For touchdown, the RL-51 would be able to throttle down to 40% thrust. This lowered the the thrust to weight ratio enough that, while the vehicle certainly couldn't hover, with some very precise timing, it would be able to perform a “suicide burn” and touchdown safely on one engine. Everything would have to go right for the landing to be successful.
Everyone involved the program that had also worked in the X-29 knew that this was a whole 'nother ballpark. The first hop took place on November 23, reaching an altitude of just 100 meters. Seeing the massive stage rise into the air was awe inspiring for all witnessing. And watching come down was even more impressive. The landing system performed correctly, with the vehicle soft landing on its exhaust flame, like something straight out of Flash Gordon. After landing, the vehicle was recovered and returned to the Vehicle Processing Facility, a specially built facility nearby where it would be inspected and refurbished. After 43 days, the vehicle was returned to the test site for a second hop. This time, the vehicle reached 3 kilometers, once again touching down perfectly. This flight used the reaction control thrusters much more than the first test, and to test vehicle maneuverability, involved horizontal translation. The vehicle was purposefully steered away from the landing pad, over the ocean, so that the booster would have to correct itself in order to land safely. Over the next six months, the vehicle would perform six additional hop tests, reaching up to 20 kilometers, and introducing significant horizontal movement. Between each flight the vehicle was carefully inspected, with ground crews learning from every flight. They managed to lower turnaround time, even with more inspections than planned for the actual flight article, to an average of 29 days. After the success of the hop tests, for the eighth flight test, the vehicle was moved to the actual launch pad first time. This flight would be much closer to the actual launch profile, with the vehicle reaching supersonic velocities, and traveling far down range. Flight 8 lifted off on June 15, 1990. Unlike the hop tests, where the fuel tanks had only been filled to a small fraction of full capacity, on flight 8 the vehicle would carry a little over half the nominal fuel load. It flew far out over the Pacific Ocean, fading from view for observers. It traveled hundreds of kilometers downrange, and reached a maximum altitude of 70 kilometers. During the flight, the booster would experience similar aerodynamic and thermal stresses to an actual launch. It successfully performed the boost back burn, and oriented itself for landing. Unlike every other test, flight 8 started with its landing gear retracted, and extended them as it approached the pad. However, despite the rest of the flight going perfectly, as PB-01T came in to land, it suffered a sticky gimbal on the center engine, which led to it coming down with too much horizontal velocity. The huge booster crashed into the pad and exploded, creating a sizable crater.
Even with the spectacular ending, the flight accomplished many of its goals, proving that vehicle could survive the flight and return to the launch site. In fact, many engineers were impressed that the vehicle even hit the launchpad at all. Thankfully, a failure at some point had been planned for, and a second test vehicle, booster PB-02T. Compared to the first vehicle, PB-02T was closer to an actual flight booster. Unlike PB-01T, which had been built purposefully overweight, with much wider margins, the second booster was very close to the final dry mass of the finished product. The landing gear, which on the first booster had used disposable crushable foam, now had actual shock absorbing hardware. This was a bit heavier, but much more reusable, as the previous landing gear had to basically be completely disassembled after every flight. Also, the new booster was equipped with a set of small aerodynamic fins Midway up the vehicle, to better steer it during descent and reentry. Most notably however, instead of beginning its career with short hops from the landing pad, PB-02T would be lifting off from the launch pad from the very start. The first flight of the new booster occurred on September 3, 1990. It basically flew a repeat of the first booster’s final flight. However, this time, it stuck the landing.
Additional flights followed, in November, December, and January, each pushing the envelope a little farther. On the fifth flight on February 4, 1991, the booster crossed the Karman line for the first time, becoming the first vehicle to travel into space and propulsively land back on Earth. The booster repeated this feat on March 17, becoming the first reusable spacecraft. On April 3, for its seventh flight, the booster looked markedly different. The nose cone used for test flights normally was gone, replaced instead with an interstage, which allowed a boilerplate centaur upper stage and payload fairing to be carried. This would better approximate an actual launch, both in terms of mass and aerodynamic loads. The seventh flight would also mark the first time that the booster would lift off with a full fuel load. Flight seven went smoothly, with the booster jettisoning the boilerplate stage after main engine cutoff, simulating how an actual flight would go. While the centaur mass simulator would plunge into the Pacific Ocean, PB-02T would successfully return to the launch site, touching down on the pad. Prometheus seemed to be getting closer and closer to an actual launch system.
Meanwhile, Lockheed continued their own work on the X-31. Though the X-31 had initially been the favorite child in the ALDV programs, the X-29’s unexpected success had pushed the flyback booster to the outskirts, its funding slashed, with seemingly no future. Lockheed continued work on the vehicle though, sinking more of their own resources into it. The continued to promote their “StarClipper” design, which they had refined even further. By the time the second X-31 prototype flew for the first time in 1987, StarClipper was a fully defined proposal, incredibly fleshed out, if still just on paper. The final StarClipper design would use a massive winged booster, its main body 8 meters in diameter, powered by 5 RL-51s. It would use the same S-IVC upper stage as the Saturn II, but Lockheed had extensively studied a winged reusable upper stage for delivering cargo, or even crew. Unlike the medium class Prometheus, which could only deliver 6-7 metric tons to Low Earth Orbit, and around 2 mt to Geosynchronous Transfer Orbit, the baseline StarClipper would be a full heavy lift vehicle, capable of throwing 23 mt to LEO, and 7 mt to GTO. This would allow it to completely replace both the Saturn II and the Titan III, at perhaps ⅓ to ½ the cost. However, developing the massive StarClipper booster would a monumental undertaking, with a huge price tag, and no one seemed willing to foot the bill. The government, already working on Prometheus, did not see the point in developing a duplicate system, even one more capable. Reusable launch systems, for all their promise, had yet to prove that they could substantially reduce the cost of access to space, and if it turned out that they didn't, spending billions to develop two of them would seem rather foolish. Lockheed could not afford to develop it independently, and even if they tried to market it commercially, it might take decades to recoup the development costs. In addition, the technical challenges in building the StarClipper booster were substantially greater than those involved with the Prometheus booster. This was not just because StarClipper was a larger vehicle, but because it basically needed to fill two roles. In addition to being a reusable rocket, the StarClipper would also need to be an aircraft the size of a 747 that would fly faster than the L-2000. In an attempt to save the design, Lockheed studied the idea of converting the Saturn II first stage into a flyback booster. Because of the higher density of the kerosene fuel, this vehicle would be smaller while still offering similar performance to the baseline StarClipper. However, this design too was still a huge undertaking, and had its own complications, particularly in the way that carbon soot from the Kerosene fuel made reuse of the RS-27A engines much more difficult than was possible with hydrogen engines.
And so despite the second X-31 vehicle making more than a dozen test flights by 1989, it seemed to be destined to be little more than a curious museum piece. And so, Lockheed, looking to recoup the hundreds of millions of their own money that they had poured into the program, began looking for other applications. At first they offered the X-31 to the DoD and NASA as research vehicle. The DoD didn't show any interest, and while NASA did, the agency had only modest plans for the vehicle, and this alone would not help pay back the cost of development. And so Lockheed looked to their other options. There was no way that a massive booster derived from the X-31 could be developed. But the X-31 on its own was a capable vehicle, and perhaps could be used without significant modification. It was to this end that Lockheed began to explore the idea of developing the X-31 into a commercial launch vehicle in late 1989. Lockheed had pioneered the field of commercial launch services by flying commsats to GTO aboard Titan IIIs. They began to study how to turn the X-31, with minimal modifications, into an affordable system for getting payloads to orbit. Thus, the Lockheed “StarBooster” concept was born. StarBooster would involve taking the basic X-31, scaling it up slightly, and replacing the J-2SL with an RL-51. For an upper stage, Lockheed contracted with Thiokol to build an solid rocket upper stage. This stage, known as the Castor 30, had its roots in the Minuteman missile, and would mass in at 14,000 kg. By combining the Castor 30 with the booster, the StarBooster system could loft more than two metric tons into a sun synchronous orbit. For heavier or higher energy payloads, a Star 48 third stage could be used. By using lower cost solid rockets for the expendable upper stages, instead of an expensive Centaur, Lockheed hoped that they could offer a low cost launch vehicle for the lower end of the market. StarBooster was perfectly sized for rapid and cheap delivery of a single GPS satellite. It was also optimal for launching Low Earth Orbit communications satellites, like those for the planned Iridium array, being able to launch two or three satellites per flight. It could also lift NASA orbital payloads, and potentially even student payloads. With the Titan III becoming less and less profitable, due to the rising costs and more stringent regulations surrounding its hypergolic propellants, Lockheed hoped that the StarBooster would help them recapture the launch market.
After completing the contracted run of test flights for the military, Lockheed continued flying the X-31, in order to help develop the StarBooster. On one flight, the X-31 reached an altitude of 85 kilometers, very close to the edge of space. The vehicle was proving much more reusable this time around, with crews getting turnaround times down to just over 30 days. The issues with reuse involving the J-2 engine were ironed out, and inspection between launches was reduced to a minimum. Lockheed began construction on the first StarBooster in 1989. They hoped to operate two boosters, one for each coast, each flying maybe up to five or six times a year. X-31 tests continued flying, with each iteration helping develop the program. Despite being based on the X-31, the StarBooster would need to be more advanced, lighter, capable of withstanding higher temperatures, and flying farther distances, smarter, and improved in a little of other ways. Lockheed was determined however. The government showed some interest in StarBooster, and put forward a small amount of funding for its development, which certainly helped Lockheed, especially when higher level executives were considering ditching the program. The first prototype StarBooster first flew in October of 1992, and soon commenced a test program similar to the X-31. Though there were some inside Lockheed that pushed for an accelerated program that could potentially beat Prometheus to being the first reusable launch vehicle. However this was quickly rejected, as Prometheus had an advantage both in funding, and in schedule, and the engineers behind StarBooster did not want to rush face first into failure. With the StarBooster prototype flying the first propulsive tests with an upper stage mass simulator by spring of 1993, the launcher was projected to make its first orbital launch by early 1995.
The Prometheus test program was rapidly approaching its final phase throughout the last half of 1991. After flight seven, all launches carried boilerplate upper stages, and if all went according to plan, an orbital test launch was planned by flight 10, hopefully before the end of the year. Flight eight flew in late May, essentially as a repeat of the previous launch, with a successful landing. On the penultimate test flight, flight nine, on June 15, 1991, hopes were high. The ascent portion of the flight went beautifully, just as planned. However, while the return was uneventful, as the vehicle came in for final landing, one of the landing gear failed to lock in place. This caused the booster to tip over after touching down. Thankfully, it did not explode, but the test article was damaged beyond repair. A subsequent investigation would conclude that the cause of the incident was a failure of several bolts inside the landing leg, likely due to fatigue. Perhaps this was something fixable that had been missed during the rush to improve turnaround times? The investigation also noted that due to a minute mistiming on the booster's engine throttle, the touchdown had been harder than any other landing (excepting of course the fireworks display that ended PB-01T), which combined with the fatigue issue, caused the leg to fail on touchdown. The booster was judged damaged beyond repair, and would never fly again, ending up on display in the national air and space museum. There were delays as PB-03, which had been planned as the first operational booster, was prepared for the remaining test flights. Construction on the vehicle had been 95% complete when the failure had occurred. And so, all resources were diverted to bringing the booster online to complete the test program.
In the months between flights, the DoD announced that for all Prometheus launches from the East Coast, Launch Complex 39, previously used for the Saturn V and it's successor, but now dormant, would be used. Necessary modifications to the facilities would begin construction by the end of 1991. In addition to the modifications to the pad and mobile launch structure, a landing pad and a hangar for refurbishing stages would be built. At least two boosters would operate from each coast, with more being added if necessary to meet demand. Concurrently with this, the DoD also announced, to the surprise of many, that a deal had been reached with Grumman, and the Prometheus system would be made available to NASA for their launch purposes. There were even talks of, down the road, allowing the system to be used to launch commercial payloads. However, there was a prohibition on launching any foreign satellites on Prometheus. Whether this all would entail boosters being shared or separate boosters for military and civilian payloads remained unknown.
Also during the gap, Grumman released a report in which they estimated the Prometheus boosters to be capable of reuse for up to 50 flights. With each booster being planned for 3-5 flights a year, this meant a 10+ year life span for each. Routine inspections would occur after each flight, and full scale refurbishments would occur every five flights. The final report also contained the final revised estimates of Prometheus's capability. The Centaur had been upgraded, from the RL-10s to the fuel capacity, and the new vehicle was capable of more than early estimates. Launching from Vandenberg, it was capable of placing 6,508 kg into a 300 kilometer Sun-synchronous orbit. From KSC, the vehicle could place up to 8,070 kg into a 200 kilometer low Earth orbit. To higher orbits the vehicle was quite capable as well. Prometheus was unique in that it was limited by its need to return to the launch site. On a typical trajectory used to launch payloads to high energy orbits, the booster would be too far from the launch site, and traveling too fast to return and land. To solve this problem, there were some proposals to have a “downrange landing site” either a barge or a stationary platform that the booster could land on, this saving the fuel needed to return to the launch site. However, in the interest of schedule and budget, Grumman opted to instead use a third stage for these high energy payloads. Initially they chose the older Agena stage for this purpose. Though this would increase per flight costs, it would greatly expand the capabilities of the system to a variety of orbits. With this combination, Prometheus could launch up to 3,350 kg into the transfer orbit required for GPS satellites. It could also throw up to 3,042 kg to a Geosynchronous transfer orbit. Grumman also collaborated with Lockheed Martin, maker of the Centaur stage, to design a small hydrogen powered upper stage to eventually replace the aging Agena. This stage, named the “Fawn”, would mass in at 8 metric tons, and allow up to 4,500 kg to be delivered to GTO, which would allow some satellites to be “dual-launched”. However, Fawn did not receive any government funding for the moment, as those in charge just wanted to get the vehicle online.
Despite all the increases in performance, the numbers everyone reading the report wanted to know were those associated with cost. Prometheus would be capable, performance wise, of replacing the Delta, Atlas, and unboosted Titans, and came close to matching the GTO performance of boosted Titans. However, the main question was if it would be able to beat those systems on its price tag, by a large enough margin to justify its development costs. Ultimately, the total costs per flight were unknowable until the vehicle was in operation. Various agencies, government and private, had made their own estimates on the cost per flight of Prometheus, and come up with a diverse spectrum of numbers. Some estimates put Prometheus as a launch vehicle with no significant cost reduction, while others predicted that it could lower the cost per kg by half. For their own estimates, Grumman first took the cost of building a single Prometheus booster, $600 million*, and divided it by the predicted 50 flights, coming out to $12 million per flight. Grumman had collaborated with Lockheed to bring down the cost of the Centaur, but each still added $45 million to the cost of every launch. The real nitty gritty details came down to the refurbishment costs and the per-flight costs, like the launch pad, fuel, ground crew, integration, etc. Grumman roughly estimated a total cost per flight for Prometheus to be about $130 million, with the Agena adding another $15 million for missions to GTO. Compared to the Titan IIIM, which could lift 17,000 kg to LEO for $400 million ($23,529 per kg), or the Atlas G, which could launch 3,630 kg for $125 million ($34,435 per kg), Prometheus had a cost per kilo to LEO of $18,587. These estimates vindicated those who had predicted that the partially reusable Prometheus would reduce launch costs, but not to a revolutionary degree. However, despite all these guesses and calculations, the true price tag per launch of Prometheus was yet to be proven, and would have to wait until the vehicle was operational.
Finally, in late November of 1991, PB-03 arrived at the launch site. After several weeks of check out and assembly, the vehicle sat on the pad on the morning of January 20, ready for its inaugural test flight. At 11:31 PST, the vehicle lifted off from the pad and rocketed into the sky. This test was to be similar to those performed by the previous booster, and it went perfectly, releasing the mass simulator, returning the launch site, and sticking the landing. After a successful repeat on March 11, engineers pushed for the orbital test flight to be flown. Originally the booster had two more tests planned, but they argued that the two previous boosters had proven all they needed to with their own test flights. In addition, all mishaps involving Prometheus had been in the landing phase of flight. During the ascent phases it had performed perfectly, without any incident. Prometheus had proved itself as a launcher. If anything went wrong during recovery, the payload would still reach orbit. And so it was decided that flight three of PB-03 would be the first orbital test. After months of preparation, the vehicle was mated with the first live Centaur stage and payload fairing, and rolled out to the pad. For this demo flight, the payload would be USA-112, a prototype of the satellite design planned for the next GPS series, Block IIR. On July 14, 1992, Prometheus lifted off on its first operational flight. Like all previous flights, the ascent went perfectly, and after main engine cut off, the centaur with USA-112 separated, and after a brief coast, ignited to push the payload to orbit. While the upper stage was still performing its burn, the booster reoriented itself and reignited its central engine, cancelling its horizontal velocity and sending back towards the launch site. The booster drifted down towards the pad, its fins and thrusters keeping it on course. Finally, seemingly at the last possible second, the engine ignited a third time for the landing burn, bringing PB-03 down for a safe landing. Uproarious applause broke out among all in attendance. Even though many of them had witnessed countless landings before, this time it was for real! Just minutes after the booster touched down, the Centaur burned out, successfully placing the payload into orbit. Cheers erupted once more. The first reusable launch vehicle was operational! The future was here!
*All dollar values here will use 2017 dollars, for the sake of simplicity.