Fresh on the heels of the Lunar Prospector and Clementine missions, which discovered sizable water ice in portions of the lunar surface, came a rather creative idea to make more measurements with a two-for-one spacecraft launch.

When President George W. Bush tried to give the American space program a needed focus in the wake of the loss of OV-102 Columbia and the imminent retirement of the Space Shuttle fleet he initiated, the moon became a renewed center of attention. A new space project was forming up, using Shuttle-derived technology, with a return to manned, extended lunar surface exploration. A series of new space probes were spun up as well.

Good times until the next administration chopped much of the manned project to pieces, flushing six years of preparation down the budgetary toilet in concert with an indifferent Congress.

Back on topic…

A new lunar polar orbiting mapping satellite, the Lunar Reconnaissance Orbiter, was going to fly on one of United Launch Alliance’s spunky Delta II rockets. But LRO’s spinning third stage looked to become a stability problem for the Delta, so LRO was upgraded to an Atlas V, which could also lift 1,000 pounds (453 kg) more than required.

Far be it for NASA to waste a ride opportunity with the capable Atlas V and its Centaur second stage, so the folks thought of adding a secondary payload. Nineteen in-house proposals were considered, from which four reached a finalist stage. The winning selection had to be lightweight and designed to use a new adapter ring that allowed two satellites to be carried within a single payload adapter.

The Lunar Crater Observation and Sensing Satellite, or LCROSS, was chosen.

After launch, the Centaur would boost the LRO into a translunar trajectory for rendezvous and orbit four days later and begin its own chapter of impressive lunar mapping (including some mythbusting pictures of the Apollo lunar landing sites).

LCROSS would take a different, slower route. It had one quick but fascinating mission: To determine if there was water ice sitting at the poles of the moon where the surface was in permanent shadow, and measure it.

But LCROSS wasn’t some ordinary orbiter or flyby mission. To pick up some velocity and to get the proper lunar phase, lighting angle and target on the southern pole of the moon, LCROSS (still attached to the now-spent Centaur stage) performed a tongue-tying LGALRO (lunar gravity lunar assist orbit), flying by the moon a few times to use the speed to adjust its path. The process would take four months or so.

So why was LCROSS still attached to the old Centaur? Because now, to LCROSS, the Centaur was the payload.

LCROSS would use the Centaur as a projectile.

Image converted using ifftoany
LCROSS jettisons the spent Centaur stage to begin the probe’s brief but fascinating mission. (NASA)

As with the fate of several Apollo lunar-bound Saturn V S-IVBs, LCROSS would release the Centaur to smack the moon, digging a 90 feet (27 m) grave for itself while throwing out a cloud of lunar stuff that LCROSS would study for four minutes, searching for water.

Why just four minutes of data? Because LCROSS had no way to stop following the vaporized Centaur and would also plow into the moon itself, concluding its mission about as fast as the active portion of data gathering began.

LCROSS nearly didn’t make it to perform its mission. Project Manager Dan Andrews recounted how LCROSS was designed with a bunch of thrusters that formed the attitude control system, managed by an inertial measurement unit (IMU) as well as a star tracker to guide the “shepherding spacecraft” (the active LCROSS probe, now pushing the Centaur).

The team learned late in design that the star tracker’s ability to also make computed changes in velocity would work great as a backup to the IMU to keep the spacecraft oriented and on-course.

A few weeks before impact, Dan and others on his team began to reacquire communications with the probe after a scheduled loss of signal as the probe swung around the lunar poles. The team’s telemetry was dire: LCROSS was firing its thrusters wildly and had spent much of its limited fuel, threatening the mission.

Andrew’s team got the probe to stop wasting fuel and learned a hard lesson. In development, LCROSS was not subjected to a popular aerospace design truism: “Test as you fly, fly as you test.” While the team appreciated the star tracker’s computed ability to help in course correction, they failed to test this on the ground before the probe flew.

As a result, when the ACS thought the the IMU was malfunctioning, the probe, by design, switched to the untested star tracker for guidance. But the star tracker’s data wasn’t a direct measurement but an estimate that Andrews dubbed “noisy” data. This noise was part of the guidance data sent to the ACS, which translated the data as having errors. The ACS then spent a buttload of fuel trying to make corrections on the ratty star tracker data.

Thankfully, the LCROSS team was able to rectify the issue from the ground, and the probe did complete its mission.

LCROSS illustrates a key element in space probe testing that is often ignored due to time pressures, failure of imagination, or lack of vision. Always attempt to simulate your spacecraft in all modes of operation to verify that what you think should work actually does work in practice. Andrews notes that the ACS/star tracker issue issue would’ve been easily discovered on ground tests that were never made.