After its mishap in 1985, the Titan 34D remained grounded until 18 April 1986 when one lifted off carrying the final KH-9 film-return reconnaissance satellite. Unfortunately, at T + 8 seconds it exploded at an altitude of 800 feet. The fragmental debris and burning propellant caused severe damage to SLC-4E and the nearby facilities. The explosion also created a toxic cloud that rose to an altitude of 8,000 feet before being blown out to sea.1 Each strap-on of a Titan 34D comprised five full segments and a half segment, a forward enclosure and an aft enclosure. Over the years, some 940 segments of this kind had been used successfully. No Titan had been lost so catastrophically in its 30-year history.2 The preliminary report on 9 June ruled out an O-ring problem like the one that had caused the loss of Challenger several months previously, and offered a number of failure modes involving 'debonding' in one of the solid rocket motors, but insufficient material was recovered to reach a definitive conclusion.3 An examination of other segments from the same batch (which had been made five years earlier, and held in storage) showed that the flaw was not age related.4 A number of motor segments were cut into 2-foot-high slices and dissected into pie shapes for inspection. The flaws discovered in these segments strongly implied that a debonding of the rubber insulation had allowed the hot gas in the motor to make contact with the steel casing, weakening it sufficiently for the 700-psi pressure to open a hole 7 inches in diameter, which caused the explosion.^6 The Air Force initiated an intensive recovery that included improving the handling procedures for the solid motors, inspecting their segments using nondestructive methods, and adding instrumentation to monitor their operation.7^
This second consecutive failure for the Titan 34D was particularly ill timed because, with the Shuttle also out of service, the Air Force had no means of dispatching its heavy satellites. With a KH-11 lost in 1985 and now a KH-9, the USA had just one operational imaging reconnnaissance satellite. This had been launched on 4 December 1984 and was almost out of propellant.9 The Titan 34D resumed service on 26 October 1987, with one from Vandenberg deploying a KH-11.10 On 29 November 1987 the first mission from Canaveral deployed a DSP early warning satellite. Some 30 minutes after the launch of the third on 2 September 1988, Air Force Secretary Edward C. Aldridge announced that the nation was "well on [its] way to assuring access to space through a robust, flexible launch capability". Unfortunately, 5 hours later when the Transtage reached the An artist's impressi°n of a DSp Mtemte. apogee of the geosynchronous transfer orbit and attempted the 110-second circularisation burn, a fractured pressurisation pipe caused its engine to misfire, stranding a Chalet-Vortex electronic intelligence-gathering satellite.11 Fortunately, the Shuttle resumed operations later in the month.
At the time of the loss of Challenger on 28 January 1986, only six Deltas remained, and on 3 May the first to be launched was lost as a result of an electrical failure that shut down its first stage at T + 71 seconds. As the Atlas and Delta both used Rocketdyne engines, commonalities in the main electronics relay boxes and wiring harnesses meant that both had to be grounded during the investigation. All of the US launch vehicles were now out of service! The Delta resumed flying on 5 September 1986. In order to meet the sudden demand, McDonnell Douglas was able to build three from spare parts. Two were made using old MB-3 Thor engines, with the more powerful Castor IVA (TX-780) strap-ons raising its performance to match that of the 3920/PAM. These hybrids (which were given the code 4925, with the '5' indicating that the Star 48 of the PAM was now considered to be a stage of the launch vehicle rather than part of the payload) dispatched Marcopolo 1 for BSB on 27 August 1989 and Insat 1D for India on 12 June 1990. The third improvised Delta was a one-off two-stage 5920 that augmented an RS-27 with the new strap-ons, and was utilised to launch the COBE satellite for NASA on 18 November 1989.
When the European Space Agency was established in 1975, it set out to develop a three-stage launch vehicle. To exploit the rate at which the Earth rotates on its axis to maximise the ability of the vehicle to put a payload into geosynchronous transfer orbit, a launch facility was built near the equator, at Kourou in French Guiana in
South America. The development of the cryogenic third stage (the first time that Europe used this technology) was protracted and the first Ariane was not launched until 24 December 1979. This was an 'all up' test with all 'live' stages, and it succeeded.12 The second test on 23 May 1980 was a failure.13 The four SEP Viking 5 engines of the first stage fired up normally, and the vehicle lifted off at T+3.3 seconds, but 1.1 seconds later the chamber pressure in one engine fluctuated at a frequency of about 1 kilohertz. This ceased after 1.6 seconds, then momentarily reappeared for a fraction of a second at T + 28 seconds, at which time the temperature in the chamber rapidly rose, the pressure fell, and the vehicle began to be subjected to a powerful rolling torque. By T+104 seconds the roll had increased to 60 degrees per second, which had the effect of interrupting the flow of propellant to the other three engines, and a few seconds later the vehicle broke up under the stress.14 Since the wreckage fell close offshore, the engine that suffered the failure was salvaged intact.15 The investigation found that there was an issue with its propellant injector. The vehicle was grounded until June 1981 while the engine was redesigned and certified.16 The two ensuing tests were successful, but on the first operational flight on 9 September 1982 the turbopump gearing in the third stage suffered damage, shutting off the engine midway through the 720-second burn to enter geosynchronous transfer orbit, losing a pair of satellites and grounding the vehicle until June 1983.17,1V9 Arianespace's policy was to offer the owner of a lost satellite priority for the launch of a replacement as soon as that became available.
The Ariane 1 configuration was designed to insert 1,500 kilograms into geosynchronous transfer orbit, but it was possible to increase this to 1,850 kilograms. Further improvements raised this capacity to 2,000 kilograms for the Ariane 2, and to 2,600 kilograms for the Ariane 3. The Ariane 3 launch on 12 September 1985 got off to an excellent start, with the second stage shutting down as planned, but some 8 seconds later the third stage failed to start, with the loss of another two satellites. The telemetry confirmed that the firing command sequence had been issued but the ignition was 0.4 second late and failed to achieve the required thrust. It was concluded that an injector valve leak had inhibited proper ignition. The vehicle was grounded once more, this time until February 198 6.20,21,22
In the aftermath of the loss of Challenger there were simply not enough American launch vehicles to
meet the demand. Arianespace, however, was in a position to expand its operations in the short term, and promptly won three commercial satellites withdrawn from the Shuttle: Aussat A3, Insat 1C and SBS 5. During the 32 months that the Shuttle was grounded, nine Ariane launches sent 16 satellites successfully on their way, but lost another one. This failure occurred on 31 May 1986, when the third stage of an Ariane 2 had a partial, unsustained ignition, followed by another ignition 0.12 second later that was above nominal pressure, after which it finally shut down. The ignition system of the HM-7B engine used one pyrotechnic cartridge to force hot gas into the injector of the combustion chamber, and another to force gas to spin-up the turbine of the turbopump. The investigation concluded that the thermal performance of the system had been weak. Since there was no evidence of a manufacturing fault, it was decided to install a more powerful igniter, to redefine the ignition sequence, and to undertake a series of high-fidelity test firings in a high-altitude chamber to recertify the engine.23,24,25 As the third stage was common, all flight operations had to be suspended, and the next launch did not occur until 16 September 1987.
The first six Block-I NAVSTAR satellites were launched between 1978 and 1980 on converted Atlas-E/F missiles to assess the feasibility of the Global Positioning System.26 The operational system was to have six sets of four satellites in 20,000-kilometre orbits inclined at 55 degrees, with the six planes evenly distributed in longitude. Establishing this system was to be a major contract for the Shuttle. By ferrying satellites up in batches of four, each flight would be able to populate one plane of the network. Even although they were not destined for geostationary orbit, such satellites required a PAM-D2 kick motor. Their deployment was to start in 1987, and the system was to be in operation within two years. The grounding of the Shuttle by the loss of Challenger in January 1986 reinforced the need for a 'mixed fleet' in which a variety of expendable rockets would complement the reusable vehicle. As a result of the 'Shuttle only' policy, McDonnell Douglas had ceased production of the Delta, and General Dynamics and Martin Marietta were about to close down the Atlas and Titan production lines. There was therefore an imperative to restart production across the range. The Air Force withdrew some government satellites from the Shuttle (including the Block-II NAVSTARs). It requested bids for expendable launchers, and on 8 August 1986 chose four contenders.27^8 In addition to an upgraded Delta from McDonnell Douglas, a variant of the Atlas-Centaur from General Dynamics, and a Titan 34D from Martin Marietta, Boeing and Hughes proposed a new vehicle that would utilise the F-1 engine that had been developed for the Saturn V.29 In January 1987 the Air Force issued McDonnell Douglas with the contract to launch the NAVSTAR satellites one at a time.
Meanwhile, the new mixed-fleet strategy had opened the door for commercial launches.30 Martin Marietta set out to compete with Arianespace, estimating that it should be able to secure 15 commercial launches between 1989 and 1994 for satellites offloaded from the Shuttle.31 In June 1987 the company announced that it intended to invest $100 million in the production of 18 Atlas-Centaurs (which it later renamed the Atlas I), three of which were booked in October by the National Oceanic and Atmospheric Administration for its geostationary weather satellites.
The first Atlas I was launched successfully on 25 June 1990, but the second, on 18 April 1991, was frustrated at T + 360 seconds when one of the two engines of the Centaur did not start. The asymmetric thrust made the vehicle tumble, and the range safety officer destroyed it. The telemetry readings were normal until one of the turbopumps refused to spin up, preventing the engine from starting. Since there was no physical evidence, the cause could not be conclusively established. The hypothesis was that a 'foreign object' - "perhaps a nut or a bolt" - had been left inside the turbopump.32 This loss of a Japanese communications satellite was written off as one of those mishaps that occur from time to time.33 Having flown over 200 missions without a failure, the Centaur had a remarkable record. Nevertheless, it was modified to yield extra torque in order to overcome a 'slow start' in a turbopump, and the software was rewritten to shut down and rerun the startup cycle in the event of an engine failing to start.34 The Atlas I resumed flying on 14 March 1992. But on 22 August 1992 the Centaur failed to start.35 It was a replay of the previous failure and since the 'foreign object' argument could not be accepted twice there was clearly a systems failure.36 The hypothesis was that the pre-launch chill down procedure had resulted in moisture from the ambient air freezing inside the turbopumps. The increased torque should have produced a successful start, but had not. The new software had ordered a second startup cycle, but without result.37 It was decided to add a solenoid valve to prevent air from entering the turbopump and freezing onto its blades.38
With the Centaur grounded again, urgently required satellites were offloaded to other launch vehicles. On the next flight on 25 March 1993, the Atlas I stranded the first of the UHF Follow-On (UFO) communications satellites for the US Navy.39 At T + 24 seconds the Atlas's two auxiliary engines started to lose thrust, levelling off at 65 per cent, which delayed the 5.5-g acceleration that triggered the jettisoning of the auxiliaries and left the sustainer with insufficient propellant to complete its burn. The Centaur extended its first firing by 24 seconds in order to attain the required trajectory, but then ran dry during the burn to enter geosynchronous transfer orbit.40^1 The investigation identified the problem as the pressure regulator for the oxygen turbopump that fed both of the auxiliaries.42 Further analysis established that an inadequately tightened screw in the regulator allowed a calibration shaft to rotate out of adjustment. The shaft was redesigned, and quality control improved.43 The good news for Pratt & Whitney was that the Centaur had ignited properly. The fault in the Atlas was readily rectified, and on 3 September 1993 a replacement UFO satellite was successfully deployed.
Was this article helpful?