The University of Michigan's solar racing team showed its mettle in the grueling World Solar Challenge marathon
Glinting in the hot afternoon sun like a giant yellow-and-black cockroach on wheels, the sleek electric racer sat on a stretch of flat vacant road on the north coast of Australia. Its eight square meters of solar cells put out barely enough power to run a hair dryer, but more than enough for the aerodynamic speedster to cruise at highway speeds. Unfortunately, at that moment, it was motionless as the driver desperately squeezed the hand throttle.
"No throttle," he said over a radio link to the chase car, where a team of engineering students from the University of Michigan sat, crestfallen.
In just 16 hours, at 8 a.m. on 18 November, the team and their car, called M-Pulse, had an appointment at the starting line of the World Solar Challenge. The race, held every two years, is an America's Cup for engineers--a grand international contest, rife with technological intrigue, that brings together many of the winners and even the mere survivors of more than a dozen grueling regional races held in the United States, Europe, and Asia. It sends state-of-the-art solar-powered electric cars 3000 kilometers across the entire Australian continent, from Darwin in the far north to Adelaide in the far south.
Solar-car racing is about the most technologically demanding pastime within reach of students and hobbyist-engineers. In fact, corporate teams from automotive giants used multimillion-dollar budgets to win two of the five Solar Challenges held before 2001. In the first race, in 1987, General Motors triumphed with Sunraycer, which ultimately evolved into the EV-1, the electric passenger vehicle introduced into select markets in 1996. And the Honda Dream, built at a reported cost of at least US $8 million, took first place in the 1996 race and set records that still stood as the 2001 race got under way.
Michigan, which had been in the Australian event three times, had never done better than a 1990 finish in third place. With M-Pulse, though, Michigan had its fastest and most reliable car in years. And with an unusually strong field of competitors, the team agreed to let IEEE Spectrum ride along in a support vehicle that would accompany M-Pulse in what promised to be an exceptional race.
To race to win, as Michigan intended to do, required a car with state-of-the-art solar cells, batteries, electronics, and motor and an elegant integration of mechanical, aerodynamic, and materials-science aspects [see "A peek under the hood"]. "There are five or six teams that have all those bases covered, and are moving beyond them," said Nader Shwayhat, the Michigan team leader, who received a master's degree in industrial engineering this past August. Besides Michigan, other pre-race favorites among the field of 33 at the 2001 event were Alpha Centauri, from the Netherlands; the Aurora Vehicle Association, from Melbourne, Australia, which had won the 1999 race; Solar Motions, a U.S. team of rowdy engineers from Silicon Valley, and the University of Missouri-Rolla.
With so many determined contenders, confidence was high that at least one of them would beat Honda's records at last.
Ups and downs Down Under
For the Michigan team, it had been a long, tumultuous, and costly journey to Darwin. To design and build M-Pulse, dozens of students had worked for two years and raised more than $1 million through donations and sponsorships. In an incredibly eventful June and July, they rebuilt the car after a devastating crash, and then beat back a tangle of technical problems to edge out Missouri-Rolla and win the American Solar Challenge, a 3700-km race from Chicago to Claremont, Calif.
The ups and downs continued after the team arrived in Australia, in mid-October, to fine-tune M-Pulse for the World event. Initially, there were weeks of rock-solid reliability, leaving plenty of time to scheme and trade gossip, the unofficial pastimes of the Challenge. A lot of the rumors swirled around the Alpha Centauri team, a newcomer, whose car, Nuna, was outfitted with solar cells of allegedly astounding efficiency and output (and cost--supposedly $1.3 million).
"We've heard crazy numbers, like 28 percent efficiency and 2100, 2200 W," said Blair Lorimer, one of four EEs on the Michigan team. The scuttlebutt invited incredulity, considering that the world record for triple-junction gallium arsenide at the time was 32.2 percent. For comparison, M-Pulse's cells were about 19 percent efficient, and those of the Aurora vehicle, the defending champion, were around 21 percent.
As Michigan nervously eyed Alpha Centauri, the Dutch neophytes returned the compliment. One afternoon, the Michigan team's Adriaan Zuiderweg came upon a small crowd of Alpha Centaurians clustered near the Michigan vehicle talking emphatically among themselves in Dutch. Zuiderweg, who was born in the Netherlands and lived there as a boy, listened for a few moments before introducing himself in flawless Dutch. The startled Alpha Centaurians ended their conversation. "They were impressed [with M-Pulse] and thought we were the team to beat," Zuiderweg translated.
Without much warning, on 14 November, the electrical glitches began. A really annoying one caused the rear-wheel steering system to spasm sporadically. On 16 November, 36 hours before race time, a group that included the team's beleaguered electrical engineers--Eric Carlson, Blair Lorimer, Chris Deline, and Saaj Shah--pulled an all-nighter to get rid of the spasms. The problem was traced to noise in a line to the servo motor that controls the rear-wheel steering.
Dawn gave way to a bright, brutally humid morning on 17 November, a Saturday. Michigan team members downed bowls of breakfast cereal or napped on sleeping bags on the concrete floor of a cavernous truck-repair garage at Darwin's Northern Territory University, where the team had set up shop. As Carlson, an IEEE student member, reprogrammed some electrically programmable ROMs, aeronautical engineering major Joe Lambert adjusted a stepper motor that controlled the rear-wheel steering. Deline, a junior and veteran of several races, forecast more trouble before the run to Adelaide began early the next morning. "I guarantee, there will be something else that will force me to stay up all night tonight."
Later that day, at a mandatory trial at a nearby motor speedway, M-Pulse hit a top speed of 112.5 km/h, the best of the 33 entrants. The top showing earned Michigan the right to lead the entire field at the next morning's start. But the team's elation over their preliminary victory would be short-lived. It was a couple of hours later, as the group prepared to make fine adjustments to the vehicle's motor, that the throttle went dead and the driver radioed the bad news to his teammates in the chase car. The preliminary assessment: a key subsystem known as the motor controller had burned out again, for the third time in three days. Deline and company looked like getting their second consecutive all-nighter after all.
That night, as a soft wet mist blanketed Darwin, the Michigan EEs worked in the yellow glare of sodium-vapor lamps in the truck garage. Chewing cold greasy slices of pizza and gulping cola, they scrutinized the entire electrical system, finding several snags that could account for their motor controller woes.
In operation, the controllers take the direct current from the car's battery pack, about 90 to 135 V, and convert it to three phases of alternating current for the car's electric motor. Keeping track of the rotor position, the motor controller fires the correct phase to keep the motor running smoothly and efficiently. The controller's two main parts are a power stage, based on an array of FETs, and a 5-V logic stage, which generates the properly timed wave signals amplified by the power stage.
Among the bugs the students found was a loose connection outside the controller circuitry that, they believed, caused a voltage spike. The spike, in turn, seemed to have damaged the 5-V supply for a logic stage, and also some of the power-stage transistors.
At 4:30 a.m., with the controller fixed, the show saved, and the last of the pizza gone, it was time to pack up M-Pulse and get it to the starting line on the other side of Darwin. As some morning clouds burned off, a throng of tourists, journalists, and curiosity seekers milled among the 33 race cars, which were lined up for several kilometers behind the starting line.
Stressed and bleary-eyed, team members nevertheless answered questions politely and posed for snapshots as a military band played off-key renditions of 1970s U.S. pop songs. Local dignitaries feted the entrants over a public address system in front of a luxury hotel. And then the cars were off. Michigan led the pack out of town, past crowds of waving spectators, solar car racing being a bigger deal in Australia than in most other places, and on to the Stuart Highway.
Gradually, Darwin dissolved into its scrubby green outskirts, with smatterings of Melaleuca, Pandanus, and Bougainvillaea trees and towering orange termite mounds. The temperature soared to over 40 °C. Huge, multisection Australian trucks called road trains thundered by, as they would throughout the race, buffeting the solar racers, which weighed several hundred kilograms at most. Inside M-Pulse, driver Jason Kramb found he was sharing the cramped, sweltering quarters with an Australian giant walking stick, a remarkably ugly bug the size of a business card.
After 300 km, with M-Pulse still in the lead, Michigan team members in a support car ahead of the racer spotted in the road the first of what would be an impressive tally of dead kangaroos before the race's end. They stopped their car and sprang into action to clear the roadkill, which could easily damage the low-slung racers. Meanwhile, Nuna, the Alpha Centauri car, was making its move: cruising at over 100 km/h, it was methodically making its way through the pack and was barely a hundred meters behind Michigan.
Moments later, south of the gritty, windswept town of Katherine, M-Pulse rolled into the first of the race's 30-minute "media" stops--just 55 meters ahead of Nuna. As journalists watched, the two teams tightened steering fixtures on their racers and sprayed the vehicles' solar arrays with water to remove dust and cool the cells to increase their efficiency. Eager alternates replaced the drivers, sweating in the pounding heat, to spell them for the next several hundred kilometers.
At 12:48 p.m., Alpha Centauri pulled ahead of Michigan and into the lead. And at 3:25 p.m., with Nuna about 14 km down the road, Michigan fell into third place as Aurora whooshed by M-pulse just before the day's second media stop, at Dunmarra. By the end of that first day, Alpha Centauri was about 15 km ahead of second-place Aurora and 29 km ahead of Michigan.
According to the rules, each racer must stop before5:10. As teams set up camp for the night, they angled their vehicle's solar array to catch as much light as possible from the setting sun. The procedure maximized the charge on the batteries.
In the Michigan camp that first evening, just north of the town of Elliott, the students trudged around in the fine red dirt, swatting at swarms of persistent flies. Lambert, Kramb, and aerospace engineering major Ashley Milne performed the nightly inspection of M-Pulse's electrical and mechanical systems; Zuiderweg checked weather forecasts; Alicia Frostic, a mechanical engineer, set up tents.
That night, as massive road trains thundered by from time to time, there was an uplifting celestial bonus, a reason to force open bloodshot eyes at 2 a.m. Amid the remote outback's spectacular profusion of stars, the Leonid meteor shower put on a dazzling show of streaking light, provoking applause from a few Michigan contestants reclining in sleeping bags on top of the team's big trailer.
With its impressive first-day showing, Nuna seemed to be living up to pre-race expectations. But like any great competitive endeavor, solar car racing is complex, with a rich strategic side that--as much as killer solar cells--ultimately decides winners and losers. "It's almost like sailing, where wind and weather play a big role in what you do, when you do it, and how you do it," explained Michigan driver Kramb. Indeed, top teams have their own weather specialists, who travel with a satellite terminal that lets them download weather maps.
"It's all about battery management," said Michael Yagley, one of two Michigan strategists. Short of a crash, a dead or badly depleted battery is the supreme disaster in solar car racing. With a battery close to spent, a team has no choice but to slow way down to let the solar array replenish it. The process could take hours, during which time the team could easily lose a hundred kilometers to competitors.
To understand solar racing, picture the vehicle as a conventional car with a tiny gas tank and terrible fuel efficiency. The objective, of course, is to keep the car going from morning to night at the highest possible average speed. The twist is that the "fuel" (electricity from the solar cells) trickles into the tank (battery) and also flows out of it at very variable rates. The weather, the car's speed, the grade of the road, even the wind all affect how many electrons flow into the battery or how many are sucked out of it. Thus the most prudent rate of speed at any given instant depends not only on the battery's state of charge, but also on what lies ahead: a breezy downhill run or a tough uphill slog, sunny skies or a rainstorm.
To deal with this variability, all the competitive teams depend on sophisticated software that simulates the entire race course. It lets team strategists pick the best possible speed at any instant given the battery's state of charge and the weather and grade ahead of the car. The battery-charge data are relayed from the racer via telemetry to a chase car, where the strategists generally sit hunched over powerful laptops running the simulation software. Periodically, the strategists radio the drivers with a target speed, which the drivers must adhere to. During the race, the stressed-out strategists rule.
Both Alpha Centauri and Michigan used strategy software based on so-called genetic algorithms, which were pioneered at Michigan in the 1970s. These find an optimal solution to a problem by mimicking, in a greatly compressed period, the mechanisms of biological evolution. The programs create a "population" of solutions and then employ techniques suggestive of mutation and crossover to "evolve" an optimal solution. According to Wubbo Ockels, a leader of the Alpha Centauri team, Alpha's strategy software was a modification of a program written at the European Space Agency, Paris, to optimize trajectories for moon launches.
In this esoteric subdiscipline, Peter J. Pudney, the Aurora strategist, is a fashion-challenged star, with his bright red socks and perpetually mussed hair. He holds a Ph.D. in solar-racing strategy, the only one in the world, earned at the University of South Australia, Mawson Lakes. The Aurora team secretary, Sorin Ionascu, called the diminutive Pudney "our little weapon. During the race, he is in charge."
That authority was apparent during the race's second day, as Aurora and Alpha Centauri dueled for the lead north of Alice Springs. At 11:46 a.m., Aurora rolled into the media stop at Tennant Creek--1 minute ahead of the Alpha Centauri car, Nuna. But not long after, as Pudney was studying telemetry data from the racer, he noticed that the readings of battery current did not jibe with other monitored readings.
It turned out the current was being undermeasured. "I started frantically going through the voltage curves" to estimate it, Pudney said. "We'd tell the driver [Adrian Marziano]: 'Coast for a while' " as they tried to get a handle on the current levels.
When Pudney realized, that afternoon, that the battery drain had been higher than he had thought, he was forced to take drastic measures. "I heard Peter say, 'Adrian, sixty' "--kilometers per hour, said Nejdeh Ghadimi, an Aurora team member who had been sitting next to Pudney in an Aurora support vehicle. "That was it--the flashing lights passed us." The flashing lights belonged to cars in the Alpha Centauri motorcade. Surprisingly enough, though, by day's end, neither Nuna nor Aurora had made it to Alice Springs, the halfway point in the race--as Honda had at the end of Day Two in 1996.
Early that evening, as fading golden sunlight fed the arrays, Ockels patiently answered a reporter's questions and cleared up the mystery of the Nuna solar cells. They were about 24 percent efficient, on average, he said. Ockels, a former astronaut with the European Space Agency (ESA), proudly pointed out a small patch of cells on the car that had actually been used in space on the Hubble Space Telescope. Retrieved by an ESA astronaut, they had been donated to the Alpha Centauri team as a kind of high-tech talisman.
The next morning, Nuna and Aurora resumed their duel, speeding along about 15 km apart, past the Tropic of Capricorn and on to the stop at Alice Springs, where they arrived before 9 a.m. Michigan followed a half-hour later, with telemetry from the racer showing that its motor was pulling way too much current. The diagnosis? A seized bearing in the motor. The cure? A motor change.
The Solar Challenge's rules prohibit changing parts at any of the 30-minute media stops, so the instant the stop ended, the team's pit crew sprang into action. Lambert barked orders as arms and legs flew and tools clanged, the swap proceeding with the swiftness and precision of a heart transplant. As the operation ended, Lambert hurled a tool to Mark Eadie, a mechanical engineering major, who snatched it out of the air and dived into a support vehicle that was already rolling to take up its position in the caravan. Total elapsed time: just over 4 minutes. The crowd of spectators at the stop yelled and cheered as the vehicles sped away in a cloud of ruddy dust.
Meanwhile, in the Aurora chase car, Pudney was still inferring the Aurora racer's battery current from its voltage. Nevertheless, the opportunity to catch Nuna was slipping away, so he took a chance and called for speeds around 100 km/h. Headwinds helped his cause--they impeded Aurora less than Nuna, which had a higher coefficient of drag. The two cars were 10 km apart early in the afternoon, just south of Alice Springs, as they sped through sunwashed groves of Mulga, spiky Acacia, gnarled pines, and the occasional Currawong tree, bright green and shiny against the baked red dirt.
On and on they cruised, just kilometers apart, past the dividing line between the states of Northern Territory and South Australia. The sun beat down, nourishing the solar speedsters chasing each other through a vast, empty flatlands with trees stubby and sparse as far as the eye could see.
By Marla Bore, 455 km south of Alice Springs, Aurora had managed to shave Nuna's lead to just 2 minutes. But with the headwind gone, Nuna's advantage in power became overwhelming. At the end of that third day of the race, Nuna was just past the desolate, opal-mining town of Coober Pedy. With the finish line just 830 km away, Nuna was 15 km ahead of Aurora and 129 km ahead of Michigan.
Still, in the Michigan camp, leader Shwayhat had not quite given up hope. "What we're hoping is that those two are competing with each other so intensely that they're draining their battery packs more than they'd like," he said as the sun set. He couldn't resist a wry smile. "That's the hope, anyway."
As Alpha Centauri set out the next morning, 21 November, greatness was within reach. The finish line was 830 km away. It would take the best single-day performance in the history of solar racing, but a finish was possible before 5:10 p.m. And that day, Alpha got everything it needed: lots of sun and no headwinds. After pushing 100 km/h for much of the morning, Nuna dashed at 110 km/h from Port Augusta to Adelaide, making the finish line at 5:09 p.m. A minute more and they would have been penalized for continuing past the 5:10 cutoff.
Alpha's overall time of 32 hours and 39 minutes beat Honda's 1996 record by 53 minutes. And their average speed of 91.81 km/h beat Honda's 1996 average of 89.76 km/h.
In the Aurora camp, Pudney had done a little calculating and discovered that his team, too, could beat Honda's records. They would need sunlight enough to provide at least 6 A h before 8 a.m. the next morning. They got exactly 6 A h and, just as Pudney predicted, Aurora slipped past the finish line at 8:43 a.m. on 22 November, beating Honda's former records by 19 minutes. Aurora's average speed was 90.26 km/h.
Michigan rolled in at 9:49 a.m. to take third place, with an overall time of 34 hours and 19 minutes and an 87.37-km/h average speed. A few kilometers past the official finish, the car paused before the ceremonial finish, as the students gathered around.
In a chase car, team leader and pop-music fan Shwayhat turned to Lambert with one last command: "Joe, put on 'I Can See Clearly Now,' and blast it." Then, with the right musical accompaniment, they sprinted across the line with their race car, hugging each other and exulting. Journalists, TV cameramen, and photographers clustered around, and Shwayhat produced champagne, which they sprayed on each other. The sun peeked out from a cloud bank.
"These kids will be at least 2 years ahead of their peers," said faculty advisor Brian E. Gilchrist, an IEEE member. "They've met deadlines and they've dealt with adversity."
Soon, the Michigan students would be going their own ways, some back to their studies, others to jobs or job searches. But for now, there was time to sip champagne and sing along one more time to the Michigan fight song, blaring on the chase car's stereo. "I can't believe it's over," said Lorimer, squinting in the late-spring sun. "I kind of wish it were still going on."
To read the updates on the 2001 World Solar Challenge filed by Sr. Editor Glenn Zorpette from Australia, see /WEBONLY/special/nov01/solar.html
To Probe Further
Several books have been written on World Solar Challenge races, including Speed of Light: The 1996 World Solar Challenge, by David M. Roche, et al. (International Specialized Book Services, 1998). The Web sites of the top finishers in the 2001 race are at http://www.alpha-centauri.nl/, http://www.aurorasolarcar.com/main_frame.html, and http://www.engin.umich.edu/solarcar/. The World Solar Challenge also maintains a site at http://www.wsc.org.au/.