Although it had been used in aircraft before, nobody had ever tried to build an entire airframe out of the material. Even drilling a hole was a problem, due to titanium's extreme hardness. Drills would be worn out after only seventeen holes. A special West German drill was found that could drill 150 holes before needing resharpening.
Before beginning production, Johnson decided to build a sample of the wing structure and nose section. When the wing structure was put in the 'hot box,' to simulate the high temperatures, it literally wrinkled. The solution was to put corrugations in the wing skin. At high temperatures, the corrugations only deepened slightly. Johnson was jokingly accused of building a Mach 3 Ford Trimotor (which also had a corrugated skin). The nose segment was used to study requirements for cooling the pilot, camera, and systems.[123]
A continuing problem during development was the poor quality of the titanium. A full 80 percent was rejected; the material was so brittle that it would shatter like glass if dropped. This problem continued into 1961, until a group from CIA headquarters went to the Titanium Metals Corporation and briefed company officials about Oxcart. The supply soon became satisfactory.[124] Lockheed also established an extensive quality-control program.
There were times, Johnson later recalled, 'when I thought we were doing nothing but making test samples.'[125]
Sometimes the problems with titanium bordered on the bizarre. During heat tests, bolt heads would simply fall off after one or two runs. It was found that cadmium plating had flaked off the tools used to tighten the bolts. This was enough to 'poison' the titanium, causing a spiderweb network of cracks to form. All cadmium-plated tools had to be thrown in a big vat that was boiling 'like a witch's brew' to strip off the plating. It was also found that welds of wing panels done during the summer soon failed, while those made during the winter lasted indefinitely. Again, it was a chemical reaction. The parts were washed before welding, and in the summer, Burbank city water had chlorine added to reduce algae. Even an ordinary pencil was dangerous. A shop worker took a pencil and wrote some numbers on a piece of titanium; a week later, it was discovered the graphite had etched the metal.[126]
Not simply the airframe, but every part would have to withstand temperatures higher than ever before endured by an aircraft. Johnson said later,
'Everything on the aircraft, from rivets and fluids up through materials and power plants, had to be invented from scratch.' All electrical connections were gold-plated, as gold retained its electrical conductivity better at high temperatures than copper or silver. The control cables were made of Elgiloy, a steel, chromium, and nickel alloy normally used in watch springs.[127] A hydraulic fluid was developed to withstand temperatures of 650 degrees F (150 degrees hotter than normal). [128]
Fuel was a difficult problem. During subsonic cruise, such as during refueling, temperatures would drop to negative-90 degrees F. At Mach 3, the fuel would be heated to 285 degrees F. It would then be pumped through the afterburner exit flaps, acting like hydraulic fluid to control their position.
This would raise its temperature to 600 degrees. The fuel would then be pumped into the J58 engine. Conventional fuel would boil and explode at such temperatures. The fuel developed was JP-7, also called LF-2A. It had a low vapor pressure; if a match was thrown into a pool of JP-7, the match would go out.[129]
The internal stress caused by such heat affected the quartz glass window for the camera. The heat had to be even throughout the window, or there would be optical distortion. This one problem took three years and $2 million to solve. The quartz window was fused to its metal frame using high-frequency sound waves.
The effect of these many problems was to delay the program and raise its cost.
Development of the J58 engines and their nacelles proved the most difficult problem. The J58 program was begun in late 1956 to power a navy attack plane with a dash speed of Mach 3. This speed would be maintained for only a few seconds. By late 1959, however, navy interest was fading, and it was decided to cancel the engine. The CIA requested the work be continued and the engine be modified for a continuous speed of Mach 3.2.
A contract was issued for three ground test and three flight test engines. [130]
With the many design changes needed to accommodate the extreme heat, virtually nothing remained of the original navy J58 engine when development was finished. To give one example, a standard ground test stand could not simulate the heat and altitude conditions required. Pratt and Whitney built a new test stand in which a J75 engine's exhaust was run through and around the J58. Speeds over Mach 3.6 and altitudes of 100,000 feet could be simulated.[131]
For all its power, the J58 engine alone was not enough to drive the A-ll to Mach 3 by brute force. The nacelles were the key that opened the way to those speeds. They were not simply a place to put the engines, but an integral part of the propulsion system. Up to 1,600 mph, air would come in through the intake and a ring of centerbody bleed vents to feed the engine.
As the A-ll approached Mach 3, the flow cycle would change. Air was now vented
It was a remarkable achievement, but years of development and flight testing would be needed before the system was reliable.
Once development began in earnest, the question became where to test the A-ll. Despite the success of the U-2 flight tests and the A-ll mock-up radar tests, Groom Lake was not initially considered. It was a 'Wild West' outpost, with primitive facilities for only 150 people. The A-ll test program would require more than ten times that number. Groom Lake's five-thousand-foot asphalt runway was both too short and unable to support the weight of the Oxcart. The fuel supply, hangar space, and shop space were all inadequate.
Instead, ten air force bases scheduled for closure were examined. (This indicates the scale of operations envisioned.) The site had to be away from any cities and military or civilian airways to prevent sightings. It also had to have good weather, the necessary housing and fuel supplies, and an eighty-five-hundred-foot runway. None of the air force bases met the security requirements, although, for a time, Edwards Air Force Base was considered.
In the end, Groom Lake was the only possibility. Plans were drawn up for the necessary facilities. As cover, the site was described as a radar test range. The remote location was explained as necessary to reduce interference from outside sources. Construction began in September 1960, several months after the CIA U-2 operation closed down. The first construction workers were housed in surplus trailers. A new water well was drilled, but the site still lacked anything but the basics.
The first major construction work was the 8,500-foot runway. This was built between September 7 and November 15, 1960, and required some 25,000 yards of concrete. This was followed by construction of the fuel storage tanks. A-ll test operations would need 500,000 gallons of JP-7 per month. By early 1962, a tank farm with a storage capability of 1,320,000 gallons was completed. Three surplus navy hangars were obtained, moved to Groom Lake, then reassembled at the north end of the facility. The navy also provided over 100 surplus housing buildings. Additional warehouse and shop space was added. Repairs to the existing buildings from the U-2 days were also made. To provide access, 18 miles of highway leading into the site were resurfaced. This work was done on a two- shift basis and continued into mid-1964.
The CIA ran into a legal problem with the construction work. Nevada law required that the names of all contractor personnel who stayed in the state for more than forty-eight hours be reported to state authorities. Listing the personnel and the companies working on the project would reveal the existence of Oxcart. The CIA general counsel discovered a loop-hole — government employees were exempt. Accordingly, all contractor personnel at Groom Lake received appointments as 'government consultants.' If any questions were raised, it could truthfully be said that only government employees worked at the site.
By August 1961, a year after work began, the basic facilities had been completed to support the initial flight
