Functional Management or Project Management?

JPL’s lunar and planetary programs developed under very different organi­zational regimes. In the lunar program, Cliff Cummings and James Burke ran the Ranger program on an academic model; Burke coordinated the ac­tivities of the subsystem engineers who worked under the technical division chiefs. JPL contracted with Hughes Aircraft Company (HAC) to design and build the Surveyor lunar lander. Surveyor lacked support from JPL, whose per­sonnel concentrated on Ranger and Mariner. Because of JPL’s neglect, HAC ran the program as it saw fit. By contrast, Robert Parks and Jack James ran the planetary program, which consisted initially of the Mariner spacecraft to fly by Venus, on the formal model they had developed on Sergeant. Al­though Mariner’s design was a modification of Ranger, the spacecraft achieved quite different results: disastrous failures on Ranger and spectacular success on Mariner. Their contrasting fates illustrate the significant influence of orga­nizational structure and processes on the technical success or failure of space­craft.

Ranger and Surveyor were intended to support NASA’s lunar program both by attaining space achievements before the Soviets did and by helping the Apollo mission. Ranger was to take close-up pictures of the lunar surface be­fore the spacecraft crashed onto the Moon and to help engineers develop spacecraft technologies for use on other programs. Ranger had an additional goal: ‘‘to seize the initiative in space exploration from the Soviets.’’ Surveyor was to perform a soft landing on the lunar surface. Conflict between scientific and engineering goals hampered both projects. Scientists desired on-board experiments, but for engineering purposes and for Apollo support, experi­ments were a nuisance. By contrast, Mariner was a purely scientific program to explore Venus and Mars, relaying photographs and scientific data back to Earth. It did not support Apollo and, consequently, had clearer mission ob­jectives.44

By late 1959, some of JPL’s managers believed that JPL needed to change its organizational structure. They thought that Pickering’s academic structure did not work well for large projects. To investigate, JPL hired management consulting firm McKinsey and Company to assess JPL’s organization. Based on the firm’s recommendations and pressure from managers like Jack James, Pickering established project-oriented Lunar and Planetary Program offices, but he maintained the authority of JPL’s functional divisions and added the Systems Division. Pickering selected Cliff Cummings to head the Lunar Pro­gram Office. Cummings, in turn, selected his protege James Burke to head Ranger.45

Burke, a Caltech mechanical engineer, had a reputation as a brilliant engi­neering researcher and technical specialist. He had an easygoing attitude with others but drove himself very hard.46 On Ranger, which began in December 1959, Burke’s mild demeanor turned out to be a handicap. The 1959 reorga­nization created project managers, but the division chiefs from Pickering’s functional organization controlled the personnel. Project managers had little authority and had to negotiate with powerful division chiefs for personnel and support. Burke did not have the authority to force division chiefs to abide by project decisions. For example, when Mariner needed personnel, division chiefs compromised Ranger by transferring some of Ranger’s most experi­enced engineers to the more glamorous Mariner. Biweekly meetings with the divisions focused on program status and scheduling, not technical problems or systems engineering.47 Burke’s project office consisted of a single deputy, and he gave the critical systems engineering tasks to the Systems Division, ad hoc committees, and technical panels.

Although JPL had developed substantial expertise in reliability on Cor­poral and Sergeant, reliability and quality assurance engineers could only advise design engineers, who could reject their advice. With many senior engineers transferred to Mariner, Ranger’s reliability suffered. Design incon­sistency resulted from continuing changes in the scientific experiments re­quested by NASA headquarters. Ranger also suffered from a requirement to sterilize components by baking them at high temperatures, which signifi­cantly reduced electronic component reliability.48

Burke’s lack of authority inside JPL was a small problem compared to his lack of authority over external organizations. JPL reported to the Office of Space Flight Programs at NASA headquarters. Air force Atlas and Agena ve­hicles were to launch Ranger, yet launch vehicles fell under the jurisdiction of the Office of Launch Vehicles at NASA headquarters, which assigned respon­sibility for Atlas and Agena to Marshall Space Flight Center (MSFC). MSFC had responsibility for, but no authority over, the air force for these vehicles. Thus, authority for the Ranger program was divided between two NASA field centers, two headquarters offices, and NASA and the air force.49

Ranger was not a priority for MSFC or for the air force. MSFC was busy designing the Saturn I rocket, a step toward von Braun’s dream of manned space flight. For the air force, NASA’s use of Atlas and Agena was secondary to developing ballistic missiles. Agena contractor Lockheed gave priority to the hundreds of upper stages slated for the air force as opposed to the nine pur­chased by NASA. NASA did not help matters, assigning responsibility for the Agena program to a headquarters-chaired committee, where Ranger was only one of several NASA Agena users. Project personnel had to work through the committee, which reported to MSFC, which in turn coordinated with the air force, which then directed Lockheed.

With this confusing organization, launch vehicle problems were a virtual certainty. Having many organizations interposed between JPL and Lockheed led to misunderstandings about the electrical and physical connections be­tween JPL’s spacecraft and Lockheed’s Agena upper stage. Exasperated JPL engineers could not get crucial Agena information from the air force or Lock­heed because they did not have the ‘‘need to know’’ required by air force security.

In September 1960 Lockheed sent a mockup of the Agena upper-stage inter­face hardware to JPL. Not surprisingly, the hardware did not match JPL’s expectations. After the ensuing investigation, the air force granted security clearances, then let NASA sign its own contract with Lockheed in February 1961. The problems also led managers and engineers in the air force and NASA to hold a design review covering interface hardware in December 1960. JPL engineers sent tooling and spacecraft mockups to Lockheed to check interface designs, so that when they manufactured flight spacecraft, they would match Agena’s interfaces.50

Mariner was JPL’s showcase project, intended to fly two spacecraft past Venus in 1962 and two more past Mars in 1964. Originally slated to launch with the new high-energy Centaur upper stage, NASA canceled the Mariner A spacecraft (the first of the series) when it became clear that Centaur would not be available in time. Regrouping, JPL engineers lightened the design to launch on Atlas-Agena launch vehicles. NASA approved the new Mariner R

spacecraft in the fall of 1961. Mariner benefited from its allure as a planetary mission and from its stable complement of onboard science experiments.51

Although Mariner’s organization included elements similar to Ranger’s, a number of features significantly strengthened Mariner’s management. As with Ranger, project managers emphasized interfaces between the spacecraft and launch vehicle, required significant testing, used JPL’s matrix structure, and had a small project office. There the similarities ended. Robert Parks, former Sergeant program manager, ran JPL’s planetary programs, and he selected his Sergeant deputy, Jack James, as project manager. The two Sergeant vet­erans decided to use Sergeant’s best management features, particularly fail­ure reporting, design freezes, and change control. Mariner engineers began by writing functional specifications to resolve spacecraft interface problems. They then created a design specification manual that defined the preliminary design, mission objectives, and design criteria. James created a development operations plan outlining the communication processes for the project, in­cluding interfaces, technical design decisions, schedule reporting, and design status meetings. James’s plan even specified what topics each status meeting should cover and who should attend. Unlike on Ranger, on Mariner James tracked the development of specifications and design drawings, not just hard­ware.52

James believed that the most innovative management feature of Mariner was the use of progressive design freezes. After a survey of subsystems to de­termine when to freeze each design element, the project periodically pub­lished a Mariner R change freeze document, along with any approved changes to drawings or specifications. Once frozen, a component’s design could be modified only through an engineering change requirement form approved by James.

Problem reporting became one of the project’s significant innovations. Project manager James instituted the ‘‘P list,’’ a list of critical problems. Any problem that made the P list received immediate attention and extra re­sources. The project implemented a failure reporting system for Mariner in November 1961, starting with system integration tests for the entire spacecraft. Failure reports were distributed to division chiefs, the project office, and engi­neers responsible for designing components and subsystems.53

As JPL prepared for its first Ranger and Mariner flights, its engineers and

Подпись: Image not available.

Mariner Venus 1962, also called Mariner 2. Mariner’s success helped convince NASA to reform JPL rather than reject it. Courtesy NASA.

managers were confident that they would succeed. Even if faults occurred, five Ranger test flights and two Mariner spacecraft gave ample margin for the un­expected. Despite last-minute changes to Ranger’s science experiments and occasional testing glitches, both projects remained on schedule. The Ranger project planned to build five ‘‘Block 1’’ spacecraft, only one of which had to work properly for Ranger’s initial objectives to be met. Surely one of them would.