A fire in the rocket engine seconds after liftoff, and then a nosedive into the ocean, are not exactly what you hope for after investing $100 million and nearly four years in developing what was supposed to be an ultrasafe vehicle for cheap space launches and space tourism.

But despite just such a setback on March 24 on Kwajelein Atoll in the Pacific Ocean, Space Exploration Technologies (SpaceX) is forging ahead with plans to develop a whole new family of semireusable rockets, which it hopes will provide a competitive, low-cost way for U.S. astronauts and cargo to get to the International Space Station once the space shuttle is retired.

[Click here to view the SpaceX rocket.]

SpaceX of El Segundo, CA, is one of at least a half-dozen companies seriously competing to revolutionize space travel in much the way a few savvy entrepreneurs transformed computing in the 1970s and ’80s. The new companies are trying a variety of approaches aimed at reducing launch costs by a factor of ten. Many of the companies started out competing for the $10 million Ansari X Prize, awarded in 2004 for the first privately financed rocket to fly into space twice in two weeks. (That rocket, the suborbital SpaceShipOne, was built by Burt Rutan’s Scaled Composites, of Mojave, CA.)

But of all the companies, only SpaceX–founded by Elon Musk, who made $300 million founding and selling the Internet payment service PayPal–is building vehicles right now that are designed to go all the way into orbit, carrying commercially valuable satellites at a fraction of current prices. While SpaceX’s rocket resembles conventional rockets, every part of it is new and designed from scratch–the first entirely new orbital rocket designed in the United States in more than a decade. And it is the first privately funded liquid-fueled orbital rocket ever developed anywhere. It is being paid for almost entirely from Musk’s personal fortune.

SpaceX uses a new kerosene-oxygen engine based on those used in the U.S. Apollo program. The company is minimizing development costs by using a single-engine design for a family of rockets that can carry a variety of payloads. The smallest of these would use one engine, the middle version five engines, the largest nine engines. This large rocket, the Falcon 9, will exceed the capacity of today’s heaviest-lift U.S. rocket, the Atlas V.

The company plans to sell launches on the Falcon 9, scheduled for its first flight next year, for $27 million, compared to an estimated $120 million for the Atlas V, and many analysts think this could open up whole new markets for the launch business, including space tourism and the deployment of scientific and commercial satellites that might be impractical at today’s prices.

Most of the new breed of space launch companies are working on innovative, totally reusable rocket systems that would return under full control, with wings like the shuttle or with retro-rockets like the Apollo moon landers. But these companies are concentrating for now on suborbital missions for tourists, which are much less demanding than actually getting to orbit.

SpaceX, which intends to go orbital from the start, does plan to make its rockets mostly reusable, but its designs are much more directly based on tried-and-true rocket designs of the past, with the addition of parachute systems to allow the individual stages to be recovered and refurbished after use. (For the Falcon 1, only the first stage is recovered.)

But this creates challenges in the development process: there is a limited amount of testing that can be done short of a full-blown launch attempt. With fully reusable flyback launch vehicles, on the other hand, it is possible to run flight tests incrementally, as X Prize winner Rutan did.

Over the course of a year, Rutan’s suborbital SpaceShipOne flew several times, gradually increasing its speed and altitude, before it made its first attempt to reach space, and many lessons were learned and adjustments made during that process.

But with expendable rockets or passive-return rockets like SpaceX’s, “you can test subsystems, but ultimately you have to test it all at once,” said Henry Vanderbilt, founder of the Space Access Society, a group advocating the development of low-cost launchers. With such systems, Vanderbilt said, the slightest mishap can be catastrophic: “Any failure tends to leave a big hole in the dirt.”

In Falcon 1’s case, it was a small fuel leak on the engine, of unknown cause. This leak caused a fire, which burned through some control lines and triggered an engine shutdown. The crash is still under investigation by SpaceX and the U.S. Air Force, which paid for the satellite onboard and provided the launch facilities.

But that should not be taken as a sign that the design is faulty, Vanderbilt stressed: “If you look at the statistics of new launch vehicles, about 45 percent have had a catastrophic failure on the first flight,” he said. And that includes workhorses like Russia’s Soyuz, which has one of the best overall success rates to date in more than 30 years of flight, despite having had 12 failures in its first 21 flights.

Musk points to a bright side. “I actually consider this first launch a partial success, because we were able to test so much hardware working together in flight, as well as empirically verify the payload environment (vibration and acoustics are worst during the liftoff phase),” he says. “The reason we started with Falcon 1 was specifically to test out the critical technologies at a small scale and without people on board before flying large vehicles with people. All complex technology developments are fraught with difficulty. Even the space shuttle has had two failures, despite almost a hundred billion dollars spent and tens of thousands of people working on it. It makes a lot more sense to work out the problems at a small scale with low-cost satellites than at a large scale with people.”

Nevertheless, since SpaceX’s more traditional approach to rocket development is different from many other new companies’, it will continue to be a focus of analysis. As aerospace consultant and rocket engineer Charles Lurio of Boston put it, “Sometimes taking a half-traditional, half-unconventional path creates a key wedge in upsetting an establishment applecart. But sometimes the conventional part of such a venture erodes one’s ability to do that.” Which is it with SpaceX? “It’s far too early to tell,” Lurio said.

Lurio emphasizes that SpaceX has done everything possible, within the confines of its chosen design, to follow the new-space mantra of “build a little, test a little.” “I’ve praised their patient testing on the ground,” Lurio said, “but it’s inescapable that to test an expendable in flight…is to lose everything if anything goes really wrong.”

That presents a daunting challenge for an engineer. “It’s like building a car new in every detail–engine, brakes, transmission, all of it,” Lurio says, “and requiring that the first time it actually engages gears and moves out of the garage, it flawlessly operates in every way over a transcontinental trip.”

And yet one of the paradoxes of this new business is that this inherently risky approach is seen by many as the “safer” path, because reusability, which allows for slow, incremental testing, is itself such a new concept in rocketry that it is seen–and for good reason–as an unproven innovation.

Musk himself remains undaunted by the initial launch failure. “SpaceX is in this for the long haul, and come hell or high water, we are going to make this work,” he said. He hopes to launch Falcon 1 again within six months and remains committed to flying the full-size Falcon 9, also from Kwajelein Atoll, in 2007.