NASA managers on Tuesday delayed the space shuttle Discovery's launch from May until at least July 1 while engineers replace four fuel-level sensors in the shuttle's external tank. These fuel sensors are responsible for making sure the shuttle's main engines shut down safely at the right time to get the spacecraft into its desired orbit. One of the four sensors was giving bad readings during testing at a Louisiana assembly facility. The tank is now at NASA's Kennedy Space Center in Florida, where the replacement work will be done.After debating the issue, the shuttle management team decided on Tuesday "to be conservative and take the safe route, and replace the sensors that are in the tank," shuttle program manager Wayne Hale told reporters during an afternoon news briefing at NASA's Johnson Space Center in Houston.Hale said it would take about three weeks to do the change-out in NASA's 52-story Vehicle Assembly Building. The job involves opening up the bottom of the 153-foot-long (46-meter-long) fuel tank, replacing the sensors, then closing the tank back up and replacing the foam insulation. All that work will push the schedule for Discovery's liftoff out of NASA's May 10-23 launch window and into the July 1-19 window.A similar problem with a single fuel sensors delayed Discovery's launch last summer. The problem was intermittent, and went away before engineers could figure out the cause. Finally, mission managers decided they would launch the shuttle even if the problem cropped up again, as long as the other three sensors were working properly. As it turned out, all four sensors worked normally on launch day.The fuel-level sensors are supposed to signal the fuel system to get ready for shutting down the shuttle's main engines if propellant levels get too low. For that reason, they're also known as engine-cutoff sensors or ECO sensors.If the sensor system fails, the shuttle's propulsion system might not be able to adjust the fuel system correctly during its ascent to orbit. If the engines cut off too early, that would leave the shuttle too low to reach orbit, forcing an emergency landing. On the other hand, if the engines keep running too long, that could cause catastrophic damage to the shuttle.Wiring may have been the problem
Even after last year's launch, NASA kept trying to track down the source of the intermittent sensor glitch — and on Tuesday, Hale said investigators have found that a batch of the postcard-sized sensors had a wiring flaw involving a connection known as a swage fitting."That swage fitting in some sensors that have been removed, sometime back in the history of the program, has been noted to be a little loose, and that's caused intermittent readings of varying resistance in the sensor," Hale said. The intermittent glitch might be made worse by the jiggling that occurs while the tank is being moved around, he said.Hale said it was not yet clear whether the balky sensor in Discovery's fuel tank actually suffered that flaw, but engineers would take a closer look during the replacement operation. He said the sensor at issue was manufactured back in 1996.Hale was confident that shuttle workers would take care of the sensor change-out and other concerns in time for the July launch opportunity. "We're all very optimistic that we'll be able to wrap up the rest of our work," he said.Sensors aren't the only issue
Last summer's flight of Discovery has been the only shuttle mission flown since a chunk of foam insulation from Columbia's external tank blasted a hole in the shuttle's left wing, enabling hot gases to rip the shuttle apart during its return to earth on Feb. 1, 2003. That accident killed Columbia's crew of seven and grounded the rest of the shuttle fleet for more than two years.

Discovery's flight was designed to test the design changes and safety measures put in place after Columbia's loss. Two minutes after its liftoff last July 26, a camera mounted on the spacecraft showed a one-pound piece of foam breaking free from the shuttle's tank. The foam did no harm, but NASA again grounded the shuttles until the problem could be solved. Engineers removed the strip of foam that caused last year's problem, but the redesigned tank still has to be cleared for flight after a series of aerodynamic wind tunnel tests. NASA is dealing with other issues as well, including fixing damage to the shuttle's robotic arm that was done during the preparations for launch.Hale said the extra time would provide more breathing space for shuttle workers to deal with these other issues, and cut down on the amount of overtime that would have been required to meet the May launch schedule.NASA needs to get its aging space shuttle fleet back in orbit to complete building the international space station. In 2010, the agency hopes to retire the shuttle fleet and turn its attention to developing a new rocket and spaceship system for returning Americans to the moon.Weighing safety and the schedule
Discovery's mission to the space station would serve as a second post-Columbia test flight, delivering supplies as well as a new station crew member, German astronaut Thomas Reiter.Before Tuesday's postponement, NASA officials were hoping to perform three shuttle missions this year, including two space station assembly missions. Hale said it was still possible to meet that schedule. "I definitely would not take that off the table," he told reporters.However, Hale said the flight schedule would be determined by safety concerns rather than the calendar. He also said the mission after Discovery's — due to be flown by Atlantis in August or September — would have to be launched during daylight hours so mission managers could get yet another good look at any foam-shedding from the external fuel tank. That could reduce NASA's flexibility for launching Atlantis.In a statement, the chairman of the House Science Committee said he applauded NASA's decision to delay Discovery's mission, known as STS-121. "NASA has done exactly the right thing in pushing back the target date for the launch of STS-121 to further address safety concerns," said Rep. Sherwood Boehlert, R-N.Y.

An investigation after last summer's sensor trouble showed that there may be a manufacturing problem that causes a loose connection between the 10-year-old sensors and wires, but further testing is needed to reach a conclusive answer, Hale said.