Hang a Right at Jupiter
For space navigators, the best course to a distant object is never a straight line.
- By Michael Milstein
- Air & Space magazine, January 2001
NASA/JHU Applied Physics Laboratory
Bob Farquhar feels lucky. And that’s good, since there’s nothing he or anyone else can do now but hope for the best. His spacecraft is out there on its own, 119 million miles away, and whatever’s going to happen next has already been programmed into the onboard computers.
Farquhar, whose mild manner seems more like that of a high school teacher than a space explorer, watches and waits from an unlikely place—not the Jet Propulsion Laboratory in Pasadena, California, headquarters for almost all past U.S. interplanetary missions, but a nondescript building at Johns Hopkins University’s Applied Physics Laboratory, outside Baltimore. It could be any office park in the country, except that the room in which Farquhar sits—at the head of a long conference table—is linked to NASA’s Deep Space Network. At the moment, one of the network’s giant dish antennas is relaying signals from a boxy little spacecraft called NEAR Shoemaker in orbit around a potato-shaped asteroid known as Eros.
NEAR stands for Near Earth Asteroid Rendezvous, the first spacecraft ever to match orbits with an asteroid and hang around for an extended study of its chemical and physical makeup (“Shoemaker” was added to the name in memory of the late astrogeologist Eugene Shoemaker). While Farquhar and his team monitor the signals, NEAR orbits 62 miles above the slowly tumbling asteroid’s surface. The satellite is about to go in for a closer look, firing thrusters to cut the orbital altitude in half.
Farquhar has just learned that the engine burn, the instructions for which were long ago loaded into the spacecraft’s computers, should last 144 seconds, nudging the craft from its piddling 5 mph to a whopping 6.5 mph. “Oh, that’s great!” he exclaims, drawing quizzical looks from others watching a screen full of numbers charting NEAR’s position. “144 is 12 squared. 12 is a lucky number. I was born September 12. [My first space mission] launched 12 minutes and 12 seconds after the hour and can you guess what the date was?
Farquhar’s faith in lucky numbers should not be easily dismissed, considering that he and his colleagues have sent spacecraft where no one thought possible, on less fuel and in less time than most people would have guessed. He calls himself an astrodynamicist, but he’s the unofficial king of the space navigators, a cadre of behind-the-scenes engineers who direct shiny, expensive spacecraft from here to there, with here being Earth and there being an asteroid, comet, planet, or moon.
Space navigation is like threading a needle, only a thousand times harder. With today’s space missions aiming at ever smaller targets (like asteroids), the eye of the needle gets so narrow, with so little room for error, that those threading it either succeed spectacularly, like Farquhar and his team have done so far at Eros, or they miss.
And in space, there’s no way to miss but spectacularly.
That’s what happened to the Mars Climate Orbiter, which vanished in September 1999 as it prepared to enter Mars orbit because of a now-infamous confusion between metric and English units—and more to the point, because it wasn’t where the navigators thought it was. The incident forced JPL space navigators into an unfamiliar and uncomfortable spotlight. It also proved—the hard way—that accurate navigation is every bit as vital to space exploration as raw rocket power.
In the early days of space travel, accuracy was relative. Mission planners were more than happy in the mid-1970s to put a Viking spacecraft within 25 to 30 miles of its planned orbit around Mars, recalls Myles Standish Jr., who left Yale’s astronomy department for JPL in 1972 to work on the Viking project. Tall and debonair, Standish, a distant relation to the Mayflower captain of the same name, is one of the few top JPL navigators who doesn’t belong to the “Texas Mafia,” a contingent of University of Texas astronomy and engineering graduates who seem to dominate the laboratory’s navigation and mission design section.
In those days, when Mars was being reconnoitered for the first time, the scientific goals and the aiming requirements weren’t as exacting as they are now. Today’s follow-up missions demand high-resolution photography and pinpoint targeting. And, says Michael Watkins, chief of the navigation section, “We’re doing it faster, and we’re trying to do it with smaller spacecraft and less fuel.”
Navigating in space is not fundamentally different from taking a road trip on Earth. First you need a map. Then you plot your course, decide what kind of vehicle to take, and calculate how much time and fuel you’ll need. Once en route, you compare your actual progress to this plan and adjust as necessary. All the while, you have to prepare for contingencies: What if we miss this turn? What if we use more fuel than we’d planned to?
Given that everything in space is moving constantly, the spacecraft navigator faces a last, even more devilish problem: What if, upon arrival, the destination turns out not to be where we thought it was?
Squirreled away in his office at JPL, Standish works at keeping such surprises to a minimum. His computer-generated ephemerides—which list the past, present, and future positions of all nine planets, the moon, and the sun—amount to a combination map and train schedule for meeting up with any large body in the solar system. Since planets move in predictable patterns, pinning down their past locations helps Standish plot their future positions. His ephemerides extend back as far as 3000 B.C. and forward to A.D. 3000. They’re calculated from an eclectic mix of sources—everything from telescope observations by Galileo to records of eclipses in ancient Babylon.
More is at stake than just a spacecraft missing its target. Mission planners also want to make sure that some long-forgotten probe sailing through the void a thousand years from now won’t crash into a distant planet and accidentally contaminate it with terrestrial microorganisms. “When it comes to navigation,” Standish says, “you are always trying to think of things that can happen that you wouldn’t normally think of.”
With Standish watching the planets, others at JPL track the solar system’s 67 known moons as well as its thousands of asteroids and comets. All of these objects push and pull on each other in subtle and hard-to-predict ways, and these changes throw the objects off their paths and necessitate constant recalculation of their orbits. Comets prove even more difficult to track, because the action of sunlight burns off dust and gas, which produces a rocket-like thrust powerful enough to drive the comet off course. “They just don’t behave themselves,” laments Donald Yeomans, Standish’s colleague at JPL and the man charged with mapping the travels of comets through the heavens.
Over the years, ephemerides—and therefore space navigation—have become more precise as more observations are entered into the database and new instruments are built that can measure celestial positions ever more accurately. But space navigators still accept unpredictability as a part of doing business. Because of the new precision, subtle forces of gravity and solar wind come more into play when calculating routes through the solar system. Navigators try to anticipate every tug on a spacecraft but it’s a constant struggle.
Part of what makes Farquhar so good at his business is that he was among the first to appreciate that in the new era of cost-constrained space missions, you had to substitute imagination and cleverness (free) for rocket fuel (expensive). Instead of countering gravity’s pushes and pulls, he found new ways to use them. In the 1980s he earned kudos in the space community by re-routing a little-known spacecraft called the International Sun-Earth Explorer 3—using gravity swings past the Earth-moon system—to fly through the tail of a comet called Giacobini-Zinner months before the world’s space agencies managed their own much more expensive missions to Comet Halley.
Now Farquhar is back to his old tricks, using some very delicate maneuvers to close in on Eros. Although the asteroid’s gravitational pull is so weak that a person standing on the surface could easily lift a car, it’s just enough to hold the 1,800-pound spacecraft in orbit. It’s slow going, though: If NEAR were to orbit much faster than its current few miles per hour, it would break free and fly off into space.
Sitting in the conference room, which doubles as mission control, Farquhar and his team track NEAR’s position by watching the radio signals coming back to Earth. If the signals are Doppler-shifted—like the change in pitch of an ambulance siren as it gets nearer—controllers know that gravity is accelerating the spacecraft in ways they hadn’t expected. Careful tracking is essential when dealing with an enigma like Eros: before NEAR’s arrival, nobody knew the asteroid’s exact shape or dimensions. Lacking a good map going into the encounter, engineers had to rely on tracking data—and first-time pictures coming back from NEAR’s cameras—to nail down the spacecraft’s location before committing it to its next move.
Farquhar delights in spacecraft navigation, and actually seems to relish all the complicated dips and detours. His longtime colleague Dave Dunham, who works out of a small basement office elsewhere on the rural APL campus, sifts numbers through his computers to find trajectories that offer the greatest advantage in terms of “delta-V”—a measure of the total change in spacecraft velocity, which translates roughly to the number and duration of engine burns and therefore to the amount of fuel a spacecraft has to carry. A mission that requires too much delta-V is dead before it even gets to the launch pad.
Farquhar’s team, for instance, could have launched NEAR in 1998 on a direct path to Eros, but it would have required so much delta-V that the mission would have needed an expensive Atlas rocket—prohibitive for a program with a budget cap of $150 million. The extra fuel would have left less room for scientific instruments. And there was one other fatal flaw. “Mars Pathfinder would have launched first,” Dunham says, in a tone suggesting that he would have found it unthinkable to let a JPL spacecraft take credit as NASA’s first Discovery-class mission off the launch pad.
So Dunham set his computers humming and found a more roundabout trajectory that used a gravity assist from Earth to pick up some free delta-V. It called for a launch in 1996, thereby beating Mars Pathfinder into space and using a cheaper and less powerful Delta II rocket. Looking over computer printouts of the spacecraft’s course, Farquhar noticed another dividend: With a little more delta-V, NEAR could fly past a smaller asteroid called 253 Mathilde on the way to Eros. So they added one extra engine burn—in technical parlance a “trajectory correction maneuver,” or TCM—to take the closest-ever portraits of an asteroid only one day after Farquhar’s wife’s birthday in 1997. (The Eros rendezvous, by the way, took place on Valentine’s Day 2000—just right, Farquhar figured, for an asteroid named for the Greek god of love.)
TCMs are at once a space navigator’s best friend and worst enemy. They give the mission designer control over the spacecraft’s route and the power to adjust errors in its course. But if they don’t go just right—engine burns are timed down to the second—they can introduce new errors in the trajectory. And each burn presents one more opportunity for something to go wrong, whether it’s a fuel line breaking, a valve sticking, or a tank exploding. “Every time you enable the thrusters, you’re taking a chance,” says Bobby Williams, leader of the JPL team providing navigational support for the NEAR mission, whose accent identifies him as a member of the Texas Mafia.
The NEAR team was rudely reminded of this risk in December 1998, when NEAR fired its thrusters to enter orbit around Eros. For reasons that still aren’t clear, the spacecraft went into a tumble. In what Farquhar now benignly calls an “unscheduled fuel dump,” it started spewing propellant, then shut itself off. For a tense day, the team feared the craft had been lost. In fact, it had reverted to a backup “safe mode,” aiming its solar panels at the sun to recharge its batteries, which had taken it out of contact with Earth. Controllers finally reestablished contact, but the mishap threw NEAR far off course.
Fortunately, Dunham, well-prepared navigator that he is, had devised a plan to use in the unlikely event the maneuver failed. It took another year to loop back to Eros, and a little more delta-V, but the spacecraft finally made it, getting as close as three miles from Eros this October.
“You should always have a contingency plan and a generous fuel supply,” Dunham says with a satisfied grin.
Extra fuel wouldn’t have been much help to the hapless engineers in charge of JPL’s Mars Climate Orbiter, whose story serves as the great cautionary tale of modern space navigation.
Launched in December 1998 to study the Martian atmosphere and relay signals from the Mars Polar Lander, which followed it, the Mars orbiter had an idiosyncrasy that flustered navigators: Unlike the Mars Global Surveyer that preceded it, the craft had solar panels that stuck out to one side. The lopsided design created a kind of sail that caught the solar wind, torquing the spacecraft around. Controllers had to counteract this force every day using onboard reaction wheels—spinning flywheels that could absorb the unwanted momentum. But the flywheels could store up only so much energy before they too had to be “unloaded” by a thruster firing in the opposite direction. And heavy use of the reaction wheels required the spacecraft to fire its thrusters 10 times more often than the navigators had expected.
Every time the thrusters fired, the navigators calculated the spacecraft’s change in trajectory. But because of a procedural mixup that began with a parts subcontractor, the calculations used English units instead of metric. The firings were actually more than four times stronger than they should have been, pushing the spacecraft slowly and steadily off course. “Even very small thrusts over time can really add up,” explains JPL’s Watkins.
Wandering off course should not in itself have spelled disaster. But in this case, the other essential ingredient of space navigation—precise tracking—also broke down. Navigators typically keep track of a spacecraft just as Farquhar’s team follows NEAR: by watching the Doppler shift in its signal. But this method only measures the distance in one direction: along the line of sight between Earth and the spacecraft. It does a poor job measuring the spacecraft’s motion out of that line, which unfortunately was the direction of the Mars orbiter’s error.
Had the spacecraft been carrying a camera during its final approach, it would have been obvious from the photographs of the Martian moons that the craft wasn’t where it was supposed to be. But photography wasn’t one of the mission’s scientific objectives, and cameras were left behind as an unnecessary luxury. Cost considerations also had led NASA to allow a supplementary tracking system known as VLBI (Very Long Baseline Interferometry) to fall into disuse. If the navigators had had the giant VLBI antennas on opposite sides of Earth, they could have used triangulation to fix the spacecraft’s position in three-dimensional space. But only the line-of-sight Doppler tracking was available, so mission control didn’t know the craft was off course until it was effectively too late.
“It should have been a slam dunk,” says Steven Synnott, a spacecraft imaging expert at JPL. “It should have been, but it wasn’t.”
The loss of the Mars orbiter was a crushing blow for JPL’s navigation section. As missions had grown more precise and ambitious, navigation had always kept up. Until now. “You have the impression that navigation is a floundering science,” sighs Standish, the planet-tracker. “It’s not. It’s a precise science, but you’ve got to have the right numbers.”
You also have to be honest about what you don’t know. When a team of engineers at JPL was plotting a proposed mission to Neptune several years ago, they wanted to know the planet’s exact distance from Earth so they could calculate how long the trip would take and, consequently, how much fuel they would need. Standish gave them the distance plus or minus 400 kilometers, or about 250 miles.
“They wanted it more accurate than that,” he recalls. “I said, ‘No—400 kilometers is the best I can do.’ You’ve got to plan your strategy on the fact that the distance to Neptune is not going to be known much better than that. Otherwise you’re fooling yourself.”
In the wake of the Mars Climate Orbiter loss, JPL assigned a kind of navigational SWAT team to make sure that the Mars Polar Lander did not go awry as well. Among other fixes, the team drew on a newly opened NASA checkbook to resurrect the VLBI tracking system, which helped navigators keep the lander on course for its target. In the end, they got the directions right. But an unrelated problem with the software that controlled the lander’s descent through the atmosphere led to an embarrassing crash on Mars.
Even if these kinds of disasters can be avoided, the future of space navigation may lie not in better calculations on the ground but in teaching spacecraft how to find their own way. Auto-navigation will be particularly important for small, distant targets whose coordinates aren’t well known from ground observations. Deep Space 1, a pioneering spacecraft launched in 1998 to test new spacecraft technologies and (as a bonus) encounter two asteroids, carried such a system. It was supposed to pick out its targets against a background of stars stored in onboard memory, then fly past them without any direction from the ground or independent tracking. NASA hailed the $152 million project as a success because it proved other technologies, including an ion drive engine. But that wasn’t the whole story: One asteroid proved too dim for the camera to recognize, a failure that scotched the auto-navigation experiment. To home in on your target, it helps to be able to see it.
The navigators in charge of NEAR intend to make sure they don’t lose track of their whereabouts at Eros. In a back room at JPL, astronomer William Owen uses the spacecraft’s detailed photographs of the asteroid to create computerized maps of the nearly 1,200 craters that pock Eros’ surface. Navigators will use the craters as road markers to identify where the orbiting spacecraft is and what part of the asteroid it’s looking at. Automating the system is a worthy goal for the future, but for now it would demand too much computer power, says Owen. So he maps the craters by hand. “The eyeball is still a wonderful computer,” he says.
Back at NEAR’s mission control in Maryland, Farquhar’s attention is focused on the numbers on the screen that tell him the spacecraft has begun firing its thrusters to lower its orbit. For a space navigator, this is the critical moment. Finally the numbers begin climbing, registering the signal’s gradual Doppler shift, which means the craft is dropping closer to the surface.
“This is too easy,” Dunham says.
“Well,” says Farquhar, smiling now, “it’s easy after you’ve done all the work.”
When the spacecraft hits a preset velocity, the thrusters shut themselves off. Soon the tracking data comes back: NEAR has hit its marks perfectly, and is now nestled into the desired close orbit. Farquhar and the rest of his mission team can sit back and relax. For now.