Propulsion for the Saturn Upper Stages

The initial decision to use liquid-hydrogen technology in the upper stages of the Saturn launch vehicles came from a Saturn Vehicle Team, chaired by Abe Silverstein and including other representa­tives from NASA Headquarters, the air force, the Office of Defense Research and Engineering, and the Army Ballistic Missile Agency 190 (von Braun, himself). Meeting in December 1959, this group, in­Chapter 5 fluenced by Silverstein’s convictions about the performance capa­bilities of liquid hydrogen, agreed to employ it in the Saturn upper stages. Silverstein managed to convince even von Braun, despite reservations, to take this step. But von Braun later told William Mrazek he was not greatly concerned about the difficulties of the new fuel because many Centaur launches were scheduled before the first Saturn launch with upper stages. His group could profit from what these launches revealed to solve any problems with the Saturn I upper stages.44


On April 26, 1960, NASA awarded a contract to the Douglas Aircraft Company to develop the Saturn I second stage, the S-IV. Between January and March 1961, NASA decided to use Pratt & Whitney RL10 engines in this stage. But instead of the two RL10s in Centaur, the S-IV held six such engines. Benefiting from consultations NASA arranged with Convair and Pratt & Whitney, Douglas did use a tank design similar to Convair’s, with a common bulkhead between the liquid oxygen and the liquid hydrogen. But Douglas also relied on its own experience in its use of materials and methods of manufac­ture. So the honeycomb material in the common bulkhead of the propellant tank was different from Convair’s design, drawing upon Douglas’s work with panels in aircraft wings and some earlier mis­sile designs. Douglas succeeded in making the larger tanks and S-IV

stage in time for the first launch (SA-5) of a Saturn I featuring a live second stage on January 29, 1964.45

Remarkably, this launch was successful despite a major accident only five days earlier. Douglas engineers and technicians knew that they had to take special precautions with liquid oxygen and liquid hydrogen. The latter was especially insidious because if it leaked and caught fire in the daylight, the flames were virtually invisible. Infrared TV cameras did not totally solve the problem because of the difficulty of positioning enough of them to cover every cranny where hydrogen gas might hide. So crews with protective clothes carried brooms in front of them. If a broom caught fire, hydrogen was leaking and burning.

Подпись:Despite such precautions, on January 24, 1964, at a countdown to a static test of the S-IV, the stage exploded. Fortunately, the re­sultant hydrogen fire was short-lived, and a NASA committee with Douglas Aircraft membership determined that the cause was a rup­ture of a liquid-oxygen tank resulting from the failure of two vent valves to relieve pressure that built up. The relief valves were in­capacitated by solid oxygen, which had frozen because helium gas to pressurize the oxygen tank had come from a sphere submerged in the liquid hydrogen portion of the tank. This helium was colder than the freezing point of oxygen. The pressure got so high because the primary shutoff valve for the helium failed to close when nor­mal operating pressure had developed in the oxygen tank. Testing of the shutoff valve showed that it did not work satisfactorily in cold conditions. Because this valve had previously malfunctioned, it should have been replaced by this time. In any event, Saturn proj­ect personnel did apparently change it to another design before the launch five days later. The committee “found that no single person, judgment, malfunction or event could be directly blamed for this incident," but if “test operations personnel had the proper sensitiv­ity to the situation the operation could have been safely secured" before the accident got out of hand.46

On the six test flights with the S-IV stage (SA-5 through SA-10, the last occurring July 30, 1965), it and the already tested RL10 engines worked satisfactorily. They provided 90,000 pounds of thrust and demonstrated, among other things, that liquid-hydrogen technol­ogy had matured significantly, at least when using RL10 engines.47


For the intermediate version of the Saturn launch vehicle, the Sat­urn IB, engineers for the S-IVB second stage further added to the payload capacity of the overall vehicle through reducing the weight

of the stage by some 19,800 pounds. Part of the reduction came from redesigned and smaller aerodynamic fins. Flight experience with the Saturn I also revealed that the initial design of the stage had been excessively conservative, and engineers were able to trim propellant tanks, a “spider [structural] beam," and other compo­nents as well as to remove “various tubes and brackets no longer required." But production techniques and most tooling did not change significantly.48

The S-IVB featured a totally new and much larger engine, the J-2, with more thrust than the six RL10s used on the Saturn I. This was the liquid-hydrogen/liquid-oxygen engine the Silverstein commit­tee had recommended for the Saturn upper stages on December 15,

1959, following which NASA requested proposals from industry to design and build it. There were five companies competing for the contract, with the three top candidates being North Ameri­can Aviation’s Rocketdyne Division, Aerojet, and Pratt & Whit­ney. Having built the RL10, Pratt & Whitney might seem to have been the logical choice, but even though NASA’s source evaluation

192 board had judged all three firms as capable of providing a satisfac- Chapter 5 tory engine, Pratt & Whitney’s proposal cost more than twice those of Aerojet and Rocketdyne. Rocketdyne’s bid was lower than Aero­jet’s, based on an assumption of less testing time, but even if the testing times were equalized, it appeared that Rocketdyne’s cost was still lower. Thus, on May 31, 1960, Glennan decided to negoti­ate with Rocketdyne for a contract to design and build the engine. The von Braun group and Rocketdyne then worked together on the design of the engine. A final contract signed on September 10,

1960, stated that the engine would ensure “maximum safety for manned flight" while using a conservative design to speed up development.49

Rocketdyne began the development of the J-2 on September 1, 1960, with a computer simulation to assist with the configuration. Most of the work took place at the division’s main facility at Ca – noga Park in northwestern Los Angeles, with firing and other tests at the Santa Susana Field Laboratory in the nearby mountains. By early November, the Rocketdyne engineers had designed a full – scale injector and by November 11 had conducted static tests of it in an experimental engine. Rocketdyne also built a large vacuum chamber to simulate engine firings in space. By the end of 1961, it was evident that the J-2 would provide power for not only the sec­ond stage of Saturn IB but the second and third stages of the Saturn V (then known as the Saturn C-5). In the second stage of Saturn V, there would be a cluster of five J-2s; on the S-IVB second stage of

Saturn IB and the S-IVB third stage of Saturn V, there would be a single J-2.50

Подпись:Rocketdyne’s engineers borrowed technology from Pratt & Whitney’s RL10, but since the J-2 (with its initial design goal of 200,000 pounds of thrust at altitude) was so much larger than the 15,000-pound RL10, designers first tried flat-faced copper injectors similar to designs Rocketdyne was used to in its liquid-oxygen/ RP-1 engines. Heating patterns for liquid hydrogen turned out to be quite different from those for RP-1, and injectors got so hot the copper burned out. The RL10 had used a porous, concave injector of a mesh design, cooled by a flow of gaseous hydrogen, but Rock – etdyne would not adopt that approach until 1962, when Marshall engineers insisted designers visit Lewis Research Center to look at examples. Under pressure, the California engineers adopted the RL10 injector design, and problems with burnout ceased. In this instance, a contractor benefited from an established design from another firm, even if only under pressure from the customer, il­lustrating the sometimes difficult process of technology transfer. Thus, Rocketdyne avoided further need for injector design, which, in NASA’s assistant director for propulsion A. O. Tischler’s words, was still “more a black art than a science."51

Rocketdyne expertise seems to have been more effective in de­signing the combustion chamber, consisting of intricately fashioned stainless-steel cooling tubes with a chamber jacket made of Inco­nel, a nickel-chromium alloy capable of withstanding high levels of heating. Using a computer to solve a variety of equations having to do with energy, momentum, heat balance, and other factors, de­signers used liquid hydrogen to absorb the heat from combustion before it entered the injector, “heating" the fuel in the process from —423°F to a gaseous temperature of -260°F. The speed of passage through the cooling tubes varied, with adjustments to match com­puter calculations of the needs of different locations for cooling.52

Because of the low density of hydrogen and the consequent need for a higher-volume flow rate for it vis-a-vis the liquid oxygen (al­though by weight, the oxygen flowed more quickly), Rocketdyne decided to use two different types of turbopumps, each mounted on opposite sides of the thrust chamber. For the liquid oxygen, the firm used a conventional centrifugal pump of the type used for both fuel and oxidizer in the RL10. This featured a blade that forced the propellant in a direction perpendicular to the shaft of the pump. It operated at a speed of 7,902 revolutions per minute and achieved a flow rate of 2,969 gallons per minute. For the liquid hydrogen, an axial-type pump used blades operating like airplane propellers to

force the propellant in the direction of the pump’s shaft. Operating in seven stages (to one for the liquid-oxygen pump), the fuel pump ran at 26,032 revolutions per minute and sent 8,070 gallons of liq­uid hydrogen per minute to the combustion chamber. (By contrast, in terms of weight, 468 pounds of liquid oxygen to 79 pounds of liquid hydrogen per second flowed from the pumps.) A gas genera­tor provided fuel-rich gas to drive the separate turbines for the two pumps, with the flow first to the hydrogen and then to the oxygen pump. The turbine exhaust gas flowed into the main rocket nozzle for disposal and a slight addition to thrust.53

In testing the J-2, engineers experienced problems with such is­sues as insulation of the cryogenic liquid hydrogen, sealing it to avoid leaks that could produce explosions, and a phenomenon known as hydrogen embrittlement in which the hydrogen in gas­eous form caused metals to become brittle and break. To prevent this, technicians had to coat high-strength super alloys with copper or gold. Solving problems that occurred in testing often involved trial-and-error methods. Engineers and technicians never knew, 194 until after further testing, whether a given “fix" actually solved Chapter 5 a problem (or instead created a new one). Even exhaustive testing did not always discover potential problems before flights, but engi­neers always hoped to find problems in ground testing rather than flight.54

Rocketdyne completed the preliminary design for the 200,000- pound-thrust J-2 in April 1961, with the preflight readiness testing finished in 1964 and engine qualification, in 1965. The engine was gimballed for steering, and it had a restart capability, using helium stored in a separate tank within the liquid-hydrogen tank to oper­ate the pneumatic system. Soon after the 200,000-pound J-2 was qualified, Rocketdyne uprated the engine successively to 205,000, 225,000, and then 230,000 pounds of thrust at altitude. Engineers did this partly by increasing the chamber pressure. They also ad­justed the ratio of oxidizer to fuel. The 200,000-pound-thrust engine used a mixture ratio of 5:1, but the more powerful versions could adjust the mixture ratio in flight up to 5.5:1 for maximum thrust and as low as 4.5:1 for a lower thrust level. During the last portion of a flight, the valve position shifted to ensure the simultaneous emptying of the liquid oxygen and the liquid hydrogen from the propellant tanks (technically, a single tank with a common bulk­head, but referred to in the plural as if there were separate tanks). The 225,000-pound-thrust engine had replaced the 200,000-pound version on the production line by October 1966, with the 230,000- pound engine available by about September 1967. As the uprated

versions became available, Rocketdyne gradually ceased producing the lower-rated ones.55

Even with six RL10s, the S-IV stage had been only about 39.7 feet tall by 18.5 feet in diameter. To contain the single J-2 and its propellant tank, the S-IVB had to be 58.4 feet tall by 21.7 feet in di­ameter. NASA selected Douglas to modify its S-IV to accommodate the J-2 on December 21, 1961. Douglas had already designed the S-IV to have a different structure from that of the Centaur, with the latter’s steel-balloon design (to provide structural support) be­ing replaced by a self-supporting structure more in keeping with the “man-rating" that had initially been planned for Saturn I and transferred to Saturn IB, which actually would launch astronauts into orbit. This structure was made of aluminum and consisted of “skin-and-stringer" type construction.

Подпись:The propellant tank borrowed a wafflelike structure with ribs from the Thor tanks Douglas had designed. The common bulkhead between the liquid hydrogen and the liquid oxygen required only minor changes from the smaller one in the S-IV. After conferring with Convair about the external insulation used to keep the liquid hydrogen from boiling away rapidly in the Centaur, Douglas en­gineers had decided on internal insulation for the fuel tank in the S-IV. They chose woven fiberglass threads cured with polyurethane foam to form a tile that technicians shaped and installed inside the tank. This became the insulation for the S-IVB as well.56 Thus, in this case technology did not transfer between firms, but shared in­formation helped with a technical decision.

For steering the S-IVB during the firing of the J-2, Douglas had initially designed a slender actuator unit to gimbal the engine, simi­lar to devices on the firm’s aircraft landing gear. Marshall engineers said the mission required stubbier actuators. This proved to be true, leading Douglas to subcontract the work to Moog Servo Controls, Inc., of Aurora, New York, which used Marshall specifications to build the actuators. The gimballed engine could adjust the stage’s direction in pitch and yaw. For roll control during the firing of the J-2, and for attitude control in all three axes during orbital coast, an auxiliary propulsion system provided the necessary thrust.57

Although they had the same designation, the S-IVB used on the Saturn V was heavier and different in several respects from the one on the Saturn IB. As the third stage on the Saturn V, the S-IVB profited greatly from the development and testing for the Saturn IB second stage. But unlike the latter, it required an aft interstage that flared out to the greater diameter of the Saturn V plus control mechanisms to restart the engine in orbit for the burn that would

send the Apollo spacecraft on its trajectory to lunar orbit. To match with the greater girth of the S-II, the aft skirt for the third stage was heavier than the one for the S-IVB second stage. The forward skirt was heavier as well to permit a heavier payload. The auxiliary propulsion and ullage system weighed more for the third stage of the Saturn V than the comparable second stage on the IB because of increased attitude control and venting needed for the lunar mis­sions. Finally, the propulsion system was heavier for the Saturn V third stage because of the need to restart. The total additions came to some 11,000 pounds of dry weight. Whereas the first burn of the single J-2 engine would last only about 2.75 minutes to get the third stage and payload to orbital speed at about 17,500 miles per hour, the second burn would last about 5.2 minutes and would accelerate the stage and spacecraft to 24,500 miles per hour, the typical escape velocity for a lunar mission.58

On the aft skirt assembly, mounted 180 degrees apart, were two auxiliary propulsion modules. Each contained three 150-pound – thrust attitude-control engines and one 70-pound-thrust ullage – 196 control engine. Built by TRW, the attitude-control engines burned Chapter 5 a hypergolic combination of nitrogen tetroxide and monomethyl hydrazine. They used ablative cooling and provided roll control dur­ing J-2 firing and control in pitch, yaw, and roll during coast periods. The ullage-control engines, similar to those for attitude control, fired before the coast phase to ensure propellants concentrated near the aft end of their tanks. They fired again before engine restart to position propellants next to feed lines. There were also two ullage – control motors 180 degrees apart between the auxiliary propulsion modules. These motors fired after separation from the S-II stage to ensure that the propellants in the engine’s tanks were forced to the rear of the tanks before ignition of the third-stage J-2. The two motors were Thiokol TX-280s burning solid propellants to deliver about 3,390 pounds of thrust.59

Despite the relatively modest changes in the S-IVB for Saturn V, development was not problem-free. In acceptance testing of the third stage at Douglas’s Sacramento test area on January 20, 1967, the entire stage exploded. Investigation finally revealed that a he­lium storage sphere had been welded with pure titanium rather than an alloy. When it exploded, it cut propellant lines and allowed the propellants to mix, ignite, and explode, destroying the stage and adjacent structures. The human error led to revised welding specifications and procedures. Despite the late date of this mishap, the S-IVB was ready for the first Saturn V mission on November 9,

1967, when it performed its demanding mission, including restart, without notable problems.60


The S-II second stage for the Saturn V proved to be far more problem­atic than the S-IVB third stage. On September 11, 1961, NASA had selected North American Aviation to build the S-II. The division of North American that won the S-II contract was the Space and Infor­mation Systems Division (previously the Missile Division), headed by Harrison A. Storms Jr., who had managed the X-15 project. An able, articulate engineer, Storms was charismatic but mercurial. His nickname, “Stormy," reflected his personality as well as his last name. (People said that “while other men fiddle, Harrison storms.") His subordinates proudly assumed the title of Storm Troopers, but he could be abrasive, embodying what X-15 test pilot and engineer Scott Crossfield called “the wire brush school of management."61

Подпись:When Storms’s division began bidding on the S-II contract, the configuration of the stage was in flux. Early in 1961 when NASA administrator James Webb authorized Marshall to initiate contrac­tor selection, 30 aerospace firms attended a preproposal conference. There, NASA announced that the stage would contain only four J-2 engines (instead of the later five), and it would be only about 74 feet tall (compared with the later figure of 81 feet, 7 inches for the actual S-II). The projected width was 21 feet, 6 inches (rather than the later 33 feet). It still seemed imposingly large, but it was “the precision it would require [that] gave everybody the jitters—like building a locomotive to the tolerance of a Swiss watch," as Storms’s biog­rapher put it. This sort of concern whittled the number of inter­ested firms down to seven. A source evaluation board eliminated three, leaving Aerojet, Convair, Douglas, and North American to learn that they were now bidding on a stage enlarged to at least a diameter of 26 feet, 9 inches—still well short of the final diameter. Also still missing was precise information about configuration of the stages above the S-II. The Marshall procurement officer did em­phasize that an important ingredient in NASA’s selection would be “efficient management."62

Once Storms’s division won the contract for the stage, it did not take long for NASA to arrive at the decision, announced Janu­ary 10, 1962, that the S-II would hold five J-2 engines. Designers decided to go with a single tank for the liquid hydrogen and liquid oxygen with a common bulkhead between them, like the design for Douglas’s much smaller common tank for the S-IVB. (The S-II

contained 260,000 gallons of liquid hydrogen and 83,000 gallons of liquid oxygen to 63,000 and 20,000 gallons, respectively, in the S-IVB.) As with the Douglas stage, common parlance referred to each segment as if it were a separate tank. Obviously, the common bulkhead was much larger in the second than the third stage (with a diameter of 33 rather than 21.75 feet), requiring unusual precision in the welding to preclude leakage. The bulkhead consisted of the top of the liquid-oxygen tank, a sheet of honeycombed phenolic insula­tion bonded to the metal beneath it, and the bottom of the liquid – hydrogen tank. Careful fitting, verified by ultrasonography, ensured complete bonding and the absence of gaps. Not only did fit have to be perfect but there were complex curvatures and a change in thick­ness from a maximum of about 5 inches in the center to somewhat less at the periphery.63

Unlike Douglas but like Convair (in the Centaur), North Ameri­can decided to use external insulation, which (it argued) increased the strength of the tank because of the extreme cold inside the tank, which was imparted to the tank walls. Initially, Storms’s en – 198 gineers tried insulation panels, but the bonding failed repeatedly Chapter 5 during testing. Using trial-and-error engineering, designers turned to spraying insulation directly onto the tank, allowing it to cure, and then adjusting it to the proper dimensions. Once the tanks were formed and cleaned, North American installed slosh baffles inside the tanks.64

The reason that insulation on the outside of the liquid-hydrogen tank increased its strength was the use of an aluminum alloy desig­nated 2014 T6 as the material for the S-II tanks. Employed long be­fore on the Ford Trimotor, it had the unusual characteristic of get­ting stronger as it got colder. At -400°F, it was 50 percent stronger than at room temperature. With the insulation on the outside, this material provided a real advantage with the -423°F liquid hydrogen inside. Both the oxidizer and fuel tank walls could be 30 percent thinner than with another material.

Unfortunately, aluminum 2014 T6 was difficult to weld with al­most 104 feet of circumference. On the first try at attaching two cylinders to one another, welders got about four-fifths around the circle when the remaining portion of the metal “ballooned out of shape from the heat buildup." The Storm Troopers had to resort to powerful automated welding equipment to do the job. Each ring to be welded had to be held in place by a huge precision jig with about 15,000 adjustment screws around the circumference, each less than an inch from the next. A mammoth turntable rotated the seam through fixed weld heads with microscopic precision. A huge

clean room allowed the humidity to be kept at 30 percent. In all of this, Marshall’s experience with welding, including that for the S-IC stage, helped Storms’s people solve their problems.65

Подпись:Despite such help, there was considerable friction between Storms’s division, on the one hand, and Marshall on the other, espe­cially with Eberhard Rees, von Braun’s deputy director for technical matters. North American fell behind schedule and had increasing technical and other problems. Marshall officials began to complain about management problems with the contractor, including a fail­ure to integrate engineering, budgeting, manufacturing, testing, and quality control. At the same time, Storms’s division was the victim of its own delays on the Apollo spacecraft it was also building. The weight of Apollo payloads kept increasing. This required lightening the launch-vehicle stages to compensate. The logical place to do so was the S-IVB stage, because a pound reduced there had the same effect as 4 or 5 pounds taken off the S-II (or 14 pounds from the S-IB). This resulted from the lower stages having to lift the upper ones plus themselves. But the S-IVB, used on the Saturn IB, was already in production, so designers had to make reductions in the thickness and strength of the structural members in the S-II.66

By mid-1964, the S-II insulation was still a problem. Then in Oc­tober 1964, burst tests showed that weld strength was lower than expected. On October 28, a rupture of the aft bulkhead for an S-II occurred during hydrostatic testing. As the date for launch of the first Saturn V (1967) approached, von Braun proposed eliminating a test vehicle to get the program back on schedule. Sam Phillips agreed. Instead of a dynamic as well as a structural test vehicle, the structural stage would do double duty.

But on September 29, 1965, the combined structural and dynamic test vehicle underwent hydraulic testing at Seal Beach. While the tanks filled with water, the vehicle was simultaneously subjected to vibration, twisting, and bending to simulate flight loads. Even though the thinned structure was substantially less strong than it would have been at the colder temperatures that would have pre­vailed with liquid hydrogen in the tanks, Marshall had insisted on testing to 1.5 times the expected flight loads. At what was subse­quently determined to be 1.44 times the load limit, the welds failed and the stage broke apart with a thunderous roar as 50 tons of water cascaded through the test site. The program was short another test vehicle. Storms’s people looked at the effect on the cost of the pro­gram and concluded that to complete the program after the failure would raise the cost of the contract from the initial $581 million to roughly $1 billion.67

When Lee Atwood, president of North American, flew to Hunts­ville on October 14, Brig. Gen. Edmund O’Connor of the air force, director of Marshall’s Industrial Operations, told von Braun, “The S-II program is out of control. . . . [Management of the project at both the program level and the division level. . . has not been ef­fective." Von Braun told Atwood the S-II needed a more forceful manager than William F. Parker, quiet but technically knowledge­able, whom Storms had appointed to head the program in 1961. Von Braun apparently got Atwood’s agreement to replace Parker and put a senior manager in charge of monitoring the program.68

The day after Atwood’s visit to Huntsville, Rees flew to Hous­ton, where he met with other Apollo managers, including Phillips. The Manned Spacecraft Center was managing Storms’s programs for the Apollo spacecraft, and Houston manager Joseph Shea had complaints similar to those of Rees about Storms’s control of costs and schedules. Phillips decided to head an ad hoc fact-finding (“ti­ger") team with people from Marshall and Houston to visit North American and investigate.69

200 The team descended upon North American on November 22, Chapter 5 and on December 19, 1965, Phillips presented the findings. George Mueller had already expressed concerns to Lee Atwood about the S-II and spacecraft programs at Storms’s Space and Information Sys­tems Division. In a letter to Atwood dated December 19 he reiter­ated, “Phillips’ report has not only corroborated my concern, but has convinced me beyond doubt that the situation at S&ID requires positive and substantive actions immediately in order to meet the national objectives of the Apollo Program." After pointing to nu­merous delays and cost overruns on both the S-II and the spacecraft, Mueller wrote, “It is hard for me to understand how a company with the background and demonstrated competence of NAA could have spent 4 1/2 years and more than half a billion dollars on the S-II project and not yet have fired a stage with flight systems in op­eration." He said Sam Phillips was convinced the division could do a better job with fewer people and suggested transferring to another division groups like Information Systems that did not contribute directly to the spacecraft and S-II projects.70

A memorandum from Phillips to Mueller the day before had been even more scathing: “My people and I have completely lost confidence in NAA’s competence as an organization to do the job we have given them." He made specific recommendations for man­agement changes, including “that Harrison Storms be removed as President of S&ID. . . . [H]is leadership has failed to produce re­sults which could have and should have been produced." After as-

suring Phillips and Mueller he would do what he could to correct problems, Atwood visited Downey and was reportedly impressed by the design work. He did not replace Storms, but Stormy him­self had already placed retired air force Maj. Gen. Robert E. Greer in a position to oversee the S-II. In January 1966, Greer added the titles of vice president and program manager for the program, keep­ing Bill Parker as his deputy. Greer agreed in a later interview that there were serious problems with S-II management. He revamped the management control center to ensure more oversight and in­corporated additional meetings the Storm Troopers called “Black Saturdays," implicitly comparing them with Schriever’s meetings at the Western Development Division. However, Greer, who had served at the (renamed) Ballistic Missile Division, held them daily at first, then several times a week, not monthly. With Greer’s sys­tems management and Parker’s knowledge of the S-II, there seemed to be hope for success.71

Подпись: 201 Propulsion with Liquid Hydrogen and Oxygen, 1954-91 But setbacks continued. On May 28, 1966, in a pressure test at the Mississippi Test Facility, another S-II stage exploded. Human error was to blame for a failure to reconnect pressure-relief switches af­ter previous tests, but inspection revealed tiny cracks in the liquid – hydrogen cylinders that also turned up on other cylinders already fabricated or in production. Modification and repair occasioned more delays. But it took the Apollo fire in the command module during January 1967 and extreme pressure from Webb to cause At­wood to separate Information Systems from the Space Division (as it became), to move Storms to a staff position, and to appoint recent president of Martin Marietta William B. Bergen as head of Space Division, actually a demotion for which he volunteered from a posi­tion in which he had been Storms’s boss. Bergen’s appointment may have been more important for the redesign of the command module than for the S-II, and certainly Storms and North American were not solely to blame for the problems with either the stage or the spacecraft. But by late 1967, engineers had largely solved problems with both or had them on the way to solution.72