The audacious rescue plan that might have saved space shuttle Columbia

If we die, we want people to accept it. We are in a risky business, and we hope that if anything happens to us, it will not delay the program. The conquest of space is worth the risk of life. —Astronaut Gus Grissom, 1965

It is important to note at the outset that Columbia broke up during a phase of flight that, given the current design of the Orbiter, offered no possibility of crew survival. —Columbia Accident Investigation Board Report

At 10:39 Eastern Standard Time on January 16, 2003, space shuttle Columbia lifted off from pad 39A at the Kennedy Space Center in Florida. Just under a minute and a half later, at 81.7 seconds after launch, a chunk of insulating foam tore free from the orange external tank and smashed into the leading edge of the orbiter's left wing at a relative velocity of at least 400 miles per hour (640 km/h). Columbia continued to climb toward orbit. The foam strike was not observed live. Only after the shuttle was orbiting Earth did NASA's launch imagery review reveal that the wing had been hit. Foam strikes during launch were not uncommon events, and shuttle program managers elected not to take on-orbit images of Columbia to visually assess any potential damage. Instead, NASA's Debris Assessment Team mathematically modeled the foam strike but could not reach any definitive conclusions about the state of the shuttle's wing. The mission continued. In reality, the impact shattered at least one of the crucial reinforced carbon-carbon heat shield panels that lined the edge of the wing, leaving a large hole in the brittle ceramic material. Sixteen days later, as Columbia re-entered the atmosphere, superheated plasma entered the orbiter's structure through the hole in the wing and the shuttle began to disintegrate. At Mission Control in Houston, the flight controllers monitoring Columbia's descent began to notice erratic telemetry readings coming from the shuttle, and then all voice and data contact with the orbiter was lost. Controllers continued to hope that they were merely looking at instrumentation failures, even as evidence mounted that a catastrophic event had taken place. Finally, at 9:12 Eastern Time, re-entry Flight Director LeRoy Cain keyed his communications loop and called out a rarely heard order: "Lock the doors." It was an acknowledgement that the worst had happened; the mission was now in "contingency" mode. The control room was sealed off, and each flight controller began carefully preserving his or her console's data. Columbia was gone, and all seven of her crew had been killed. NASA refers to this most rare and catastrophic of events as an LOCV—"Loss of Crew and Vehicle."

Frozen

Columbia is lost. There are no survivors. —President George W. Bush in a national address, 14:04 EST, February 1, 2003

The world of human space flight paused—first to mourn, then to discover what had happened. Congress laid that responsibility on the combined shoulders of the Columbia Accident Investigation Board (referred to, in typical NASA acronym-dependent style, as "the CAIB" or just "CAIB," which rhymes with "Gabe"). In the months after Columbia, the CAIB stretched its investigative fingers all through NASA and its supporting contractors. My own memories of the time immediately following the accident are dominated by images of somber meetings and frantic work. I was a junior system administrator at Boeing in Houston, and because we supported the shuttle program, we had to locate and send cases and cases of backup tapes—containing everything that happened on every server in our data center during the mission—over to NASA for analysis. In August 2003, the CAIB issued its final report. Behind the direct cause of the foam strike, the report leveled damning critiques at NASA's pre- and post-launch decision-making, painting a picture of an agency dominated by milestone-obsessed middle management. That focus on narrow, group-specific work and reporting, without a complementary focus on cross-department integration and communication, contributed at least as much to the loss of the shuttle as did the foam impact. Those accusations held a faint echo of familiarity—many of them had been raised 17 years earlier by the Rogers Commission investigating Challenger's destruction. In the end, Columbia's loss ended not only lives but also careers at all levels of NASA. A number of prominent shuttle program managers were reassigned. It is likely that Columbia's destruction factored heavily into the resignation of NASA Administrator Sean O'Keefe. Many involved with the mission—including many still working at NASA—to this day struggle with post-traumatic stress and survivor's guilt. All pending shuttle missions were put on hold, and Columbia's three surviving sister ships—Discovery, Atlantis, and Endeavour—were grounded. NASA looked inward, and we wondered if the orbiters would ever fly again.

A path not taken

To put the decisions made during the flight of STS-107 into perspective, the Board asked NASA to determine if there were options for the safe return of the STS-107 crew. —Columbia Accident Investigation Board Report

That's the way events actually unfolded. But imagine an alternate timeline for the Columbia mission in which NASA quickly realized just how devastating the foam strike had been. Could the Columbia astronauts have been safely retrieved from orbit? The CAIB had the same question during the writing of its report, so it asked NASA to develop a theoretical repair and rescue plan for Columbia "based on the premise that the wing damage events during launch were recognized early during the mission." The result was a remarkable document, which appears at the end of the report as Appendix D.13. It carries the low-key title "STS-107 In-Flight Options Assessment," but the scenario it outlines would have pushed NASA to its absolute limits as the agency mounted the most dramatic space mission of all time. NASA planners had one fortuitous ace in the hole that made the whole plan possible: while Columbia's STS-107 mission was in progress, Atlantis was already undergoing preparation for flight as STS-114, scheduled for launch on March 1. As Columbia thundered into orbit, the younger shuttle was staged in Orbital Processing Facility 1 (OPF-1) at the Kennedy Space Center. Her three main engines had already been installed, but she didn't yet have a remote manipulator arm or a payload for her payload bay. Two more weeks of refurbishment and prep work remained before Atlantis would be wheeled across the space center to the enormous Vehicle Assembly Building and hoisted up for attachment to an external tank and a pair of solid rocket boosters.

So an in-orbit rescue was at least feasible—but making a shuttle ready to fly is an incredibly complicated procedure involving millions of discrete steps. In order to pull Atlantis' launch forward, mission planners had to determine which steps (if any) could be safely skipped without endangering the rescue crew.

The desperate race

The scenarios were to assume that a decision to repair or rescue the Columbia crew would be made quickly, with no regard to risk. —Columbia Accident Investigation Board Report (Appendix D.13)

But even before those decisions could be made, NASA had to make another assessment—how long did it have to mount a rescue? In tallying Columbia's supplies, NASA mission planners realized that the most pressing supply issue for the astronauts wasn't running out of something like air or water but accumulating too much of something: carbon dioxide. Weight is a precious commodity for spacecraft. Every gram of mass boosted up into orbit must be paid for with fuel, and adding fuel adds weight that must also be paid for in more fuel (this spiral of mass-begets-fuel-begets-mass is often referred to as the tyranny of the rocket equation). Rather than carrying up spare "air," spacecraft launch with a mostly fixed volume of internal air, which they recycle by adding back component gasses. The space shuttle carries supplies of liquid oxygen and liquid nitrogen, which are turned into gas and cycled into the cabin's air to maintain a 78 percent nitrogen/21 percent oxygen mixture, similar to Earth's atmosphere. The crew exhales carbon dioxide, though, and that carbon dioxide must be removed from the air. To do this, the shuttle's air is filtered through canisters filled with lithium hydroxide (LiOH), which attaches to carbon dioxide molecules to form lithium carbonate crystals (Li₂CO₃), thus sequestering the toxic carbon dioxide. These canisters are limited-use items, each containing a certain quantity of lithium hydroxide; Columbia was equipped with 69 of them. How long those 69 canisters would last proved difficult to estimate, though, because there isn't a lot of hard data on how much carbon dioxide the human body can tolerate in microgravity. Standard mission operation rules dictate that the mission be aborted if CO₂ levels rise above a partial pressure of 15 mmHg (about two percent of the cabin air's volume), and mission planners believed they could stretch Columbia's LiOH canister supply to cover a total of 30 days of mission time without breaking that CO₂ threshold. However, doing so would require the crew to spend 12 hours of each day doing as little as possible—sleeping, resting, and doing everything they could to keep their metabolic rates low.

If the crew couldn't sustain that low rate of activity, NASA flight surgeons believed that allowing the CO₂ content to rise to a partial pressure of 26.6 mmHg (about 3.5 percent cabin air volume) "would not produce any long-term effects on the health of the crewmembers." This would enable the crew to function on a more "normal" 16-hour/8-hour wake/sleep cycle, but at the cost of potential physiological deficits; headaches, fatigue, and other problems related to the high CO₂ levels would have started to manifest very quickly.

After the carbon dioxide scrubbers, the next most limited consumable was oxygen. Columbia's liquid oxygen supplies were used not only to replenish breathing gas for the crew but also to generate power in the shuttle's fuel cells (which combined oxygen with hydrogen to produce both energy and potable water). The amount of liquid oxygen on board could be stretched past the CO₂ scrubbers' 30-day mark by drastically cutting down Columbia's power draw. The remaining three consumable categories consisted of food, water, and propellant. Assuming that the crew would be moving minimally, food and water could stretch well beyond the 30-day limit imposed by the LiOH canisters. To preserve propellant, the orbiter would be placed into an attitude needing minimal fuel to maintain. Exactly when the crew of Columbia would enact these power- and oxygen-saving measures depended on a short decision tree. In the scenario we're walking through, the assumption is that NASA determined on Flight Day 2 (January 17) that the foam strike had caused some damage, followed by at least another day to gather images of Columbia using "national assets" like ground-based telescopes and other space-based sources (i.e., spy satellites) under the control of USSTRATCOM.

If that imagery positively identified damage, Columbia would immediately enter power-down mode; if the images didn't show anything conclusive, the crew would conduct an EVA (extra-vehicular activity—a spacewalk) to visually assess the damage to the wing, then power things down. In either case, Flight Day 3 would mark the start of many sleepless nights for many people.

No do-overs, no mistakes

This rescue was considered challenging but feasible. —Columbia Accident Investigation Board Report

Planning the inspection EVA would have taken most of Flight Day 4 (January 19), but the hard deadline of the lithium hydroxide canisters remained set at Flight Day 30 (February 15) regardless of what happened on the ground. Work would simultaneously have had to begin at the Kennedy Space Center to accelerate the processing of Atlantis. "Accelerate" is a prosaic word for the herculean effort that would have been needed. Activities that normally take place across weeks or months would have to happen in hours or days. Civil servants and contractors at KSC would have to begin 24/7 shift work, keeping the lights on and the process running every hour of every day, for a minimum of 21 days, to power Atlantis through checkout and make her ready to launch. Three unceasing, brutal weeks of 24/7 shift work—and that's with absolutely no margin factored in for errors or failures. The Orbital Processing Facility team, the Vehicle Assembly Building team, and the Launch Complex 39 pad team would have had to get every one of the millions of steps right, and every component of Atlantis would have had to function perfectly the very first time, or it would all be wasted.

So many things would have to happen. First, Atlantis' computers would have to be reprogrammed to accommodate the changes in the mission. Fortunately, the flight software developed for STS-114's International Space Station (ISS) rendezvous could be adapted to instead rendezvous with Columbia, though most of the specific rendezvous parameters would have to be altered. The changes would be uploaded to Atlantis' computers during the DOLILU—the Day of Launch Input Load Update, the standard last-minute software update that shuttles on the pad receive two hours prior to launch. Usually, DOLILU loads include flight control updates to accommodate the day's observed weather patterns, but this particular DOLILU load would change the entire flight profile. It would be the largest on-pad software update ever attempted. In order to push Atlantis through processing in time, a number of standard checks would have to be abandoned. The expedited OPF processing would get Atlantis into the Vehicle Assembly Building in just six days, and the 24/7 prep work would then shave an additional day off the process of getting Atlantis mated to external tank and boosters. After only four days in the Vehicle Assembly Building, one of the two Crawler-Transporters would haul Atlantis out to Launch Complex 39, where she would stage on either Pad A or Pad B on Flight Day 15—January 30.

Once on the pad, the final push to launch would begin. There would be no practice countdown for the astronauts chosen to fly the mission, nor would there be extra fuel leak tests. Prior to this launch, the shortest time a shuttle had spent on the launch pad was 14 days; the pad crews closing out Atlantis would have only 11 days to get her ready to fly. Even as workers at Cape Canaveral frantically tried to beat the clock, more work had to happen at the Johnson Space Center in Houston: Atlantis still needed a crew.

The right stuff

[I]t would be important to have a high degree of confidence in the astronautsʼ ability to quickly adapt to the micro-gravity environment. —Columbia Accident Investigation Board Report (Appendix D.13)

Columbia carried seven astronauts, who by Flight Day 15 would be halfway through their unexpectedly extended 30-day mission. This presented a problem for NASA: Atlantis would need her own crew in order to launch and rendezvous with Columbia, but space shuttles were only designed to accommodate five to seven astronauts. When Atlantis returned, she would carry not only the astronauts she launched with, but also Columbia's rescued crew—so, to minimize crowding, what was the minimal crew count Atlantis could get away with at launch? After analysis, it was determined that Atlantis would need a minimum crew of four. A two-person pilot and commander team would be required to actually fly the rendezvous and actively keep station with Columbia—which NASA estimated would mean at least eight or nine hours of manual flying (and potentially much more than that). Another two-person team would be required to don suits and perform the rescue EVA tasks—tasks which NASA would have had to design from scratch. As with every other piece of the rescue, there was no room for error, and there would be no second chances. Atlantis would therefore be launched with an all-veteran crew, with selection for the mission biased heavily toward astronauts who demonstrated fast adaptation to microgravity (there was no time to be space-sick) and high aptitude at EVA or rendezvous. The report names no names, but it does indicate that an assessment revealed a pool of nine EVA candidates, seven command candidates, and seven pilot candidates available in January 2003 whom NASA felt could have undertaken the mission. The four astronauts chosen to fly Atlantis would have faced an extraordinarily compressed training schedule—and also a tremendous amount of professional and personal pressure. The tight timeline would mean that the two Atlantis astronauts selected to actually spacewalk between the shuttles for the rescue EVA would likely be training underwater at NASA's Neutral Buoyancy Lab almost every single day of the two weeks, breaking the entire multi-hour spacewalk up into tiny component maneuvers and procedures and walking through each to commit them to memory. Simultaneously, the two astronauts selected to pilot the shuttle would have spent that time in the large motion-base simulator in Building 9 at the Johnson Space Center, working through every moment of the rendezvous, station-keeping, and landing from start to finish.

It's also certain that the media would have exerted its own tremendous pressure, attempting to thrust cameras and lights into every corner of the preparation—as much as they would be allowed to do so, anyway. "Space disaster" and "rescue mission" are golden ratings words. Clear Lake in Houston and Cape Canaveral in Florida would have been swarmed with TV trucks; the Johnson Space Center sign on historic NASA Rd 1 would likely have been a constant backdrop on TV news both local and national. And throughout the frantic weeks on the ground, Columbia's crew would wait.

Slow time

This powerdown would have supported only the most basic vehicle control and crew support and communication equipment. —Columbia Accident Investigation Board Report (Appendix D.13)

While work on the ground would proceed in a controlled frenzy, time on Columbia would lengthen and draw out in slow misery. The crew would have potentially undergone a brief flurry of activity if they needed to do an EVA to confirm the damage to the orbiter's left wing; additionally, they would have needed to maneuver Columbia into a tail-first "gravity gradient" attitude so that the Earth's pull on the shuttle's empennage would keep the orbiter's orientation fixed relative to Earth without the need to expend any propellant. After that, though, the stranded crew could do very little other than wait and try not to move or breathe too much.

The crew wouldn't even be able to watch their own rescue's TV coverage, because the orbiter would be in a tightly restricted low-power mode in order to conserve its energy. Appendix D.13 includes a description of what systems would be shut down, and among them are "all cameras, camera heaters, TV monitors, and video equipment." An oft-asked question is whether or not Columbia could have docked with the ISS, which would have had consumables to spare. There are numerous reasons why this would not have been possible, but the overriding one comes down to simple physics: Columbia would have had to execute what is known in orbital mechanics terminology as a "plane change" maneuver—applying thrust perpendicular to her orbital track to shift orbit to match the ISS' more tilted inclination. Plane change maneuvers require tremendous amounts of energy—in some cases, even more energy than was required to launch the spacecraft in the first place. Appendix D.13 dismisses the possibility of an ISS rendezvous with just two sentences:

Columbia's 39 degree orbital inclination could not have been altered to the ISS 51.6 degree inclination without approximately 12,600 ft/sec of translational capability. Columbia had 448 ft/sec of propellant available.

The crew would be playing the long game, carefully conserving resources for the burst of activity that would have to occur at the end of the mission. As previously noted, the primary consumable of consequence would be the carbon dioxide scrubbers, so rest and sleep would have been the crew's main mission. Columbia's orbital period would mean that during their quiet exile, the crew would see more than 300 sunrises break over the curving lip of the world, spread across weeks of endless drifting purgatory.

Russian roulette

This new risk to the Orbiter would weigh heavily in the decision process on launching another shuttle and crew. —Columbia Accident Investigation Board Report (Appendix D.13)

Appendix D.13 is written under the assumption that the damage to Columbia's wing was recognized and acted upon, but that is actually the first of two major assumptions underlying the rescue mission. The second assumption has its own set of enormous issues: given that Columbia was disabled by a foam strike, NASA would have to be willing to subject Atlantis to the exact same risk.

The obvious unsettling question here is whether or not there was anything NASA could do in the near term to prevent Atlantis from being disabled by the same type of foam impact—and, given the nature of the strike, the answer is "no." The foam chunk that sheared off of Columbia's external tank was part of what's called the left "bipod ramp," one of two hand-sculpted structures flanking the large bipod struts that secure the orbiter's nose to the forward part of the external tank. To form the bipod ramps, orange BX-250 insulation is sprayed over the fittings that attach the bipod to the external tank. It's allowed to dry, then it's shaved by hand into wedges that cover the fitting elements. Coupled with a layer of ablative materials atop the fittings, the foam ramps both protect the attachment points from heat during launch and also sheath them in an aerodynamic shape. And, as it turns out, bipod ramps had broken off six times before STS-107.

Much of the CAIB report is given over to discussing the specifics of the external tank's insulating foam—what it's made of, how that material performs, and how often foam has sheared off of the ET and impacted with an orbiter. What is clear from the report is that the STS-107 foam strike was not a unique event—it was a relatively common occurrence that in this particular instance happened at precisely the right (or wrong) time to cause catastrophic damage to one of the very few things on the shuttle without any form of redundancy. The Atlantis rescue mission would face the exact same vulnerability. The Appendix clearly states that in the scenario outlined, Atlantis would have to fly without any time added to the processing schedule to perform assessments, upgrades, or repairs to the external tank that had already been manufactured and assigned to Atlantis for the now-postponed STS-114 mission. The odds that tank would also shed foam in a catastrophic manner were non-zero. This is another reason why the rescue crew would have been made up of four astronauts instead of launching with more crew—to expose as few humans as possible to the risk of death.

Gravity ballet

Success Criteria: The safe return of the rescue vehicle (Atlantis) and both crews. —Columbia Accident Investigation Board Report (Appendix D.13)

There would have been three launch windows during which Atlantis would be able to reach Columbia; one at 23:09 EST on February 9 (Flight Day 25), another at 22:40 EST the next day, and a final one the day after that at 22:05 EST. Columbia would be made ready to meet her younger sibling three days prior to the first launch window. The crew, potentially suffering from the effects of carbon dioxide poisoning, would revive the shuttle's systems enough to push Columbia into a slightly elliptical higher orbit, which would give Atlantis a better set of opportunities to make the rendezvous. Any of the three launch windows would have provided a working margin to reach Columbia before the crew's supply of carbon dioxide scrubbers was exhausted, but earlier was obviously better. The first launch window provided a substantially earlier rendezvous time on February 10; the latter two windows both meant a rendezvous on February 13. The later rendezvous time would leave, at most, 36 hours of margin before Columbia could no longer support life. Weather is one of the major unknowns when planning a shuttle launch—not just at the launch site but also at the multiple places around the world that must be kept ready for an emergency landing if the orbiter needs to abort its attempt to reach orbit. The CAIB report shows that luck would have been on NASA's side here; a review of observed weather conditions on the proposed launch days showed that there was nothing happening in the atmosphere that would have hindered the launch. More worrying, though, was that all three windows opened at night. A night launch would substantially reduce NASA's ability to observe foam damage during Atlantis' flight to orbit, which was particularly ominous in light of the reason behind the rescue mission. Because of this, an additional EVA was added for Atlantis' crew after reaching Columbia—they would carefully examine Atlantis' wings and tiles for any damage.

Assuming all went well and there were no countdown delays, Atlantis would have lifted off on the evening of February 9, 2003. In order to lower the nitrogen content in their blood and be ready to don their suits as soon as possible, the two EVA crew on Atlantis would possibly have been required to breathe pure oxygen from the moment they entered the orbiter's cabin on the launch pad. As the other orbiter lifted off, Columbia's crew would have long since set a shuttle program record—they would have been in space for 25 days, eight days longer than the previous longest shuttle mission.

Atlantis would arc upward into orbit, approaching Columbia from below in what's called an "R-bar approach"—that is, an approach along an imaginary radial line connecting Columbia with the Earth's center. (Contrast this to a V-bar approach, which would be an approach along Columbia's velocity vector—that is, from the front or back, rather than the top or bottom.) Columbia would already be oriented tail-first and "upside down" relative to the ground; Atlantis would approach "right side up" beneath it. Atlantis would swing slowly up into place, each shuttle growing larger and larger in the rendezvous windows in their respective flight decks' ceilings. Finally, Atlantis would ease to a halt 20 feet (six meters) from Columbia. Atlantis would be yawed 90 degrees to Columbia, pointing at three o'clock to the older orbiter's 12 o'clock, in order to keep their vertical stabilizers from striking.

This would have been the first time two space shuttles were simultaneously orbiting, and the challenges would have been considerable. Each shuttle would have its own flight control room operating in NASA's Mission Control Center—and, with the ISS also requiring another flight control room, this would have tasked the control center to capacity (both from a perspective of technical and human resources). Moreover, Atlantis would have needed to be under constant manual control for the duration of the rendezvous, because even at a distance of 20 feet, orbital mechanics would keep the two spacecraft moving at different velocities and they would drift apart in short order, with Atlantis at the lower altitude constantly trying to race ahead of Columbia.

While Atlantis' pilot and commander settled in to trade off the task of holding the shuttle steady for nine hours, the other two crew—called "EV1" and "EV2" in the report—would already be suited and standing by in the airlock. As soon as they were given a "go," EV1 and 2 would open Atlantis' airlock and retrieve an expandable boom, with which EV2 would assist EV1 in moving across the gap to Columbia. Extra lithium hydroxide canisters would also be transferred between the two shuttles to give Columbia some breathing room—literally—along with a pair of spacesuits for Columbia's crew to wear. The Appendix notes that the suits would need to be transferred "powered up, and pressurized" to preserve their water supplies, which might have proved a somewhat unexpected sight to anyone watching the EVA through the orbiter's windows.

Two undoubtedly stir-crazy Columbia crewmembers (referred to as "CM1" and "CM2") would already be suited and waiting in Columbia's airlock to assist in the transfer from Atlantis. EV1 would parcel the supplies from Atlantis into Columbia's airlock, then assist CM1 and CM2 out of the airlock and help them negotiate the path back to Atlantis.

Shell game

There would be a number of activities that would be attempted for the first time during this conceptual inspection and rescue mission. —Columbia Accident Investigation Board Report (Appendix D.13)

From here, the complexity of the mission begins to ratchet up to maximum. Depending on how tired and compromised they were physically, CM1 and CM2 could help spell the Atlantis pilot and commander at their station-keeping exercises (assuming that CM1 and CM2 were Columbia commander Rick Husband and pilot William McCool), but the two extra space suits would be put to considerable use. Atlantis' two EVA crewmembers would remain outside, and while CM1 and CM2 were removing their suits, the two Atlantis crew would use their SAFER jet packs to check over Atlantis' tiles and leading edges for damage (Columbia lacked SAFER packs, and the inspection EVA her crew would have gone through would have involved much more strenuous techniques to clamber along the orbiter's structure and get a look at the wing). CM1 and CM2 would remove their suits and then get them ready for reuse; they would be returned by EV1 and EV2 to Columbia and stashed in the airlock, which would then be pressurized and opened. Two more Columbia crewmembers would already have donned the spare suits shuttled over earlier and thus become CM3 and CM4, and the same procedure would be repeated as with CM1 and CM2.

The report spells out a best-case scenario where the suit donning and doffing goes off without a hitch, and in that case, all the transfers could be done without stopping for a break. This would mean that Atlantis' EV1 and EV2 crewmembers would be outside for somewhere between 8.5 to nine hours in a single EVA. However, that outcome is hardly a given. Putting on a space suit is a complex procedure on the ground, in full gravity and with multiple sharp-eyed assistants helping out. Putting on a suit in Columbia's middeck, possibly while still attempting to shake off the effects of carbon dioxide poisoning, is a much dicier operation. It's made even more complicated by the fact that for each successful crew transfer, the number of helpers is reduced. It's possible that the operation would have stretched to multiple EVAs—instead of nine hours, it could have taken more than three times as long. After the first two two-person transfers, the next transfer would consist of a single person: CM5 would go across alone, with EV1 and EV2 assisting. This would be done because Columbia had a crew of seven, and one person would have to do the transfer by themselves. Leaving the last two crewmembers, CM6 and CM7, to operate as a pair at least meant that the two would have had each other's assistance in donning the well-used spacesuits. (Ars consulted a number of sources to gauge the complexity of donning spacesuits in microgravity without any assistance from unsuited crew. Though none would speak on record, the consensus is that it would involve what was universally categorized as an extremely high degree of difficulty.)

Shooting star

There would be no possibility of recovering Columbia. —Columbia Accident Investigation Board Report (Appendix D.13)

As the CAIB dispassionately notes, there was no chance of recovery for the stricken shuttle. Even if the wing could have been patched and cold-soaked and Columbia's reentry profile altered to raise the angle of attack and lower the temperature at the wing leading edge, vehicle loss was almost certain. Further, even if successful reentry were possible, the shuttle could not be landed entirely from the ground—there was no way for Mission Control to have extended the shuttle's landing gear or the air probes necessary to judge velocity once in the atmosphere. Those functions (as well as starting the shuttle's auxiliary power unit) could only be invoked by physically throwing switches in the cockpit after reentry, during approach and landing. (The remaining shuttle fleet gained the ability to land totally under ground control in 2006, with the development of the RCO IFM cable, a 28-foot (8.5-meter) braided cable that the crew could use to physically link the cockpit with the shuttle's avionics bay and patch Mission Control into the required switches.)

For Columbia, this wasn't an option. Prior to their exit, CM6 and CM7 would have had to proceed to the shuttle's flight deck and toggle a number of switches into place, giving Mission Control on the ground direct command of Columbia's guidance and maneuvering systems. CM6 and CM7 would then depart, sealing the airlock behind them and leaving Columbia to find her own way. Atlantis would back slowly away from her sister ship, and her crew of 11 would busy themselves preparing for their own crowded reentry ordeal—never before has an orbiter landed with that many people on board, and even simple things like seating would be complicated. Some crew would literally have to sit strapped to the floor during reentry. At some point over the next few hours or days, ground controllers would command Columbia to close her cargo bay doors and orient herself for what would be her final task. The shuttle would roll damaged thermal tiles to face Earth and perform a retrograde burn with the large OMS engines. Shortly after that, she would cross the entry interface. Contrary to popular belief, the heat a spacecraft faces on reentry isn't generated by simple friction but rather by ram pressure—the fast-moving shuttle compresses the air in front of it, forming a massive shock zone in which air molecules ionize and break apart. As Columbia descended, an observer on the flight deck would see the windows glow and flare with plasma. After a short time, that plasma would invade the shuttle's structure through the hole in the wing leading edge. Columbia's last act would be to brighten the sky over the South Pacific, first as one glowing star and then breaking apart into many. The remains of the oldest shuttle would pepper the surface of the Pacific, and she would be no more.

Ad astra per aspera

It should be noted that although each of the individual elements could be completed in a best-case scenario to allow a rescue mission to be attempted, the total risk of shortening training and preparation time is higher than the individual elements. —Columbia Accident Investigation Board Report (Appendix D.13)

We all love Hollywood endings, but it's difficult to envision the rescue mission coming together with the required level of perfection. For example, in researching this article, I was unable to discover the number of times a shuttle has gone through an Orbital Processing Facility, Vehicle Assembly Building, and launch pad processing flow with no errors or faults. Based on the complexity of the machine, I suspect that it has never happened before. And yet, when faced with a challenge of this magnitude and with such tremendous consequences, it's incredibly attractive to imagine NASA rising to the task. As an agency, NASA simultaneously represents the best and worst of the United States of America—it is responsible for some of the greatest engineering achievements in the history of humankind and has accomplished a long list of goals originally deemed impossible. At the same time, the agency is also crippled by a lack of direction and leadership; it has gone from being an organization capable of putting human beings on other worlds to an organization that lacks even the means to put them into low Earth orbit without assistance. Update, nine years later: When this piece was originally published in 2014, the criticisms above were current and valid. However, in the almost-decade since, the world of space flight has been utterly transformed. Today in 2023, more than half of the Expedition 68 crew circling Earth on the International Space Station were delivered there by SpaceX spacecraft carried by SpaceX launch vehicles; the James Webb Space Telescope sits at the L2 point doing incredible science; and even the long-delayed SLS rocket has had its maiden flight, propelling an uncrewed Orion spacecraft into a multi-week circumlunar journey. Unlike the years spent adrift in the earlier part of this century, it feels like NASA has now turned a very important corner, and the future of human space feels bright and hopeful for the first time in a long time. The mission to rescue Columbia represents the kind of challenge that NASA has since its beginnings demonstrated an unswerving ability to meet. There would have been a clear goal, with hard timing requirements, and the agency's massive pool of engineering talent would be empowered to accomplish the goal at any cost and without restriction. The will to win would not be lacking, but technical challenges are ignorant of will and drive—look, for example, at the oxygen tank explosion that crippled the Apollo 13 command and service module in 1970. That explosion was the result of a combination of events that occurred prior to launch, with potential blame stretching from an electrical switch subcontractor all the way to the crew itself. The error-free rescue of Columbia would have depended not just on the flawless execution of teams at all of the NASA centers but also on an unknown number of events that happened days, weeks, months, or even years in the past leading up to the mission. In the writing of this piece, I spoke with a number of current and former NASA personnel, both inside and outside the Flight Operations Directorate. All were polite, but none would talk on-record about the feasibility of the proposed Atlantis rescue mission. The formal response I received from NASA's Public Affairs Office respectfully but firmly informed me that the CAIB report is NASA's full and final statement on the matter:

From NASA’s standpoint, there is nothing further to add to the Columbia Accident Investigation Report (Chapter six and its appendices) related to the "what if" scenario of rescuing the STS-107 crew. As you are aware, it is spelled out very clearly that there would have to have been a very large number of "knowns" to have executed a rescue or repair mission for Columbia at that time. ... Beyond that, we respectfully decline any specific interviews on the subject and reference you to the CAIB report for the detailed analysis provided during the investigation of the Columbia accident.

Ultimately, Appendix D.13 is a well-informed and research-backed exercise in speculation, constructed by engineers who were intimately familiar with shuttle program operations. My telling of the rescue's story is not intended to criticize or damn NASA for its actions, nor am I attempting from a position of historical privilege to second-guess the decision-makers who to this day must live with the consequences of their choices. Columbia and her crew almost certainly could not have been rescued without too many "ifs" having fallen the other way. I can tell the story of what might have been the most awe-inspiring moment in all of human space flight, but I am profoundly unqualified to speculate beyond the boundaries of the CAIB report. It's an amazing story—but it's only a story.

The long road back

It is unlikely that launching a space vehicle will ever be as routine an undertaking as commercial air travel—certainly not in the lifetime of anybody who reads this. The scientists and engineers continually work on better ways, but if we want to continue going into outer space, we must continue to accept the risks. —Columbia Accident Investigation Board Report

It took 907 days after Columbia's destruction for NASA to return to flight. STS-114—flown by Discovery instead of Atlantis—lifted off from the Cape on July 26, 2005. I remember it very well—now as a not-so-junior system administrator, I watched helplessly as the sheer number of Boeing employees streaming the countdown and launch video from NASA TV saturated our site's Internet link, which somewhat hilariously almost caused site management to consider requesting a launch hold (Boeing's Houston office provided shuttle support, and some of those support activities needed that same Internet link to function). Fortunately, the launch was a success. Starting with STS-114, no shuttle would fly without a rescue shuttle on standby. These planned emergency flights (numbered STS-3xx) were called LON missions, for "Launch On Need." In the event of trouble on a shuttle mission, crews would rendezvous with the ISS and shelter there for up to 50 days, while the LON shuttle would be made ready to fly to retrieve them. The one exception to this was the final Hubble servicing mission, STS-125. The orbital height and inclination of the Hubble made the mission totally incompatible with an emergency ISS rendezvous in the event of trouble, so a plan based partially on the Atlantis/Columbia rescue was drafted. The STS-125 LON mission would have been dubbed STS-400. Because the ISS wasn't available, STS-400's Endeavour needed to be ready to launch on short notice; this led to the final instance of what was already a rare sight: two shuttles staged at LC-39 simultaneously.

The LON missions were never needed, and the shuttle program finished without any other significant incidents. Foam strikes were not eliminated, but post-launch analysis of each shuttle was increased. It is a virtual certainty that future NASA manned spacecraft will return to their rightful place on top of launch vehicles rather than being slung on their sides in the path of debris. NASA's culture continues to evolve. It is impossible to say at this point if the lessons of Columbia have been fully inculcated into the agency. Update, nine years later: Ars senior space editor Eric Berger weighs in on the safety and culture aftermath of Columbia in this piece, also published on the 20th anniversary of the accident. I was there for the aftermath and the return to flight, but agency-wide policy changes are things that happened far above my pay grade. The thing I remember more than anything else, the single most vivid memory of them all, is of the memorial service the Tuesday after Columbia's destruction.

February 4, 2003

This cause of exploration and discovery is not an option we choose—it is a desire written in the human heart. We are that part of creation which seeks to understand all creation. We find the best among us, send them forth into unmapped darkness, and pray they will return. They go in peace for all mankind, and all mankind is in their debt. —President George W. Bush, addressing personnel at the Johnson Space Center

We arrived at the Johnson Space Center at about 9:30am, having been told that space would be limited for the service, which was to start at noon. After a half-mile of walking and a security checkpoint, we stood in the park that stretches between Building 1 and Building 8, lost amid a sea of people. The stage and podium were far away on the other side of the grassy field, and we passed the two-and-a-half hours in uncomfortable, standing silence. After a long wait, Air Force One, trailed by three F-15s, circled on its way down to Ellington Field. The crowd swelled to its maximum just after 11. At noon, with no fanfare, President Bush and First Lady Laura Bush walked together to their place on the stage. They held hands, which stuck in my mind—even the most powerful man in the world holds hands with his wife. There was an invocation, and then words from NASA's director and the chief of the Astronaut Corps. Both paid tribute to each astronaut individually, and the Corps chief clearly had to fight to keep back tears. The president stepped to the podium next and spoke eloquently about the human spirit. The only even vaguely political words that left his mouth were ultimately topical—he said that the space program would continue. Then he, too, spoke of each astronaut individually, praising their daring and dedication. A ship's bell tolled seven times, once for each of Columbia's crew, and then four NASA T-38s flew over in the missing man formation. The jets moved in low and fast, streaking toward us in a wedge less than 250 feet off the ground. As they passed overhead, the second jet back on the left side of the formation peeled sharply upward, right as the roar of the engines hammered at us. The missing man jet arched high and straight as the formation continued onward, now with an empty spot to recognize that there were men and women who were no longer with us. I have never before witnessed anything so profoundly moving as that trio of jets hurtling low over the rest of the campus, with their missing comrade thousands of feet above and rocketing higher still. I will remember it forever. Godspeed, Columbia.