Getting into orbit is difficult, but getting back down again safely is one of the trickiest aspects of spaceflight. As space agencies and commercial players set their sights on the Moon, Mars and beyond, they need to know they can depend on a safe return.
"Entering the atmosphere at a speed of Mach 25 really makes a hell of a noise in the cabin," says Belgian astronaut Frank De Winne, talking from a training mock-up of the International Space Station at the European Astronaut Centre in Cologne, Germany. His eyes widen. "You can hear the jets firing to stabilise the vehicle, and air is screaming past, and all this happens while you're in a kind of fireball as the outside heats up to about 2,000-2,500°C."
De Winne first went to space in 2002 and then again in 2009. Both times he came home in a Russian Soyuz capsule, a tiny cone-shaped Soviet-era craft that hasn't had a significant makeover since its first development half a century ago. Crew are cramped three abreast with their legs bent, and are dropped down from the edge of space to the planet below. Gravity, the atmosphere, a parachute and solid-fuel braking engines firing just before touchdown do the rest.
Re-entry is perhaps the most terrifying and low-tech part of space travel - an experience which De Winne's colleague, European Space Agency astronaut Paolo Nespoli, likened to being in a car crash.
But right now, Soyuz is the only choice for today's astronauts if they want to get home from the ISS. Despite Russia having been a major player (and other nations' ally) in space exploration for decades, the recent political tensions have led to the West urgently seeking other ways of getting astronauts back down. The past few months have been especially busy, with Nasa, Esa and commercial players such as SpaceX successfully testing new re-entry vehicles that they hope to start using before 2020. And it's no longer just about the ISS, or its replacement, expected in the late 2020s - future missions eye landing on asteroids and comets, colonising the Moon and even getting humans to Mars.
Shuttle no more
Back in the 1950s engineers worked out that the most effective re-entry shapes were blunt, with the high-drag profile creating a kind of air cushion that keeps the hot gases generated in the atmosphere away from the vehicle itself. It was the leading design until Nasa's space shuttle broke the mould with a winged glider capable of a runway landing.
The shuttle's re-entry routine was complex. First it would fly upside down and backwards towards the upper atmosphere, firing its engines to slow into a curve down to Earth; then it would flip back the right way up, do four banking turns and then aim for a narrow flight corridor. If it came in too steep, its wings would heat up dangerously; too shallow and they'd create lift. It was an expensive, but incredibly capable spacecraft that was eventually retired in 2011, prompted not least by the 2003 Columbia disaster, in which seven astronauts died.
The accident highlighted the challenge of re-entry: damage to Columbia's left wing during launch allowed hot gases inside the structure, which then led it to break apart and be destroyed.
Engineers are developing designs for a new set of demands, and taking them to space. Test flights are the only way to learn about our complex atmosphere; the Esa's head of aerothermodynamics Jose Longo says: "Neither simulation in computer nor simulation in facilities like a wind tunnel represents reality. In the end, you need to fly."
On 11 February this year, Esa tested an estate-car sized spacecraft dubbed IXV that looks like a flying slipper. The rounded nose, flat underside and vertical rear form what's called a lifting body - a shape that can't exactly glide, but generates useful lift. Loaded with some 300 sensors, the craft shot up on a Vega rocket to an altitude of 412km - 12km higher than the ISS. After a push from the Vega upper stage, it came back down into the atmosphere at 27,000km/h, the standard speed for re-entry from low Earth orbit.
The test was a success, and the engineers brought home 200GB of data about heat, pressure and angles. At the moment of re-entry - around 90km high - the IXV ploughed along at a rather improbable-looking 45°, nose in the air, its two carbon ceramic steering flaps at the back twitching at shotgun speed to guide it left or right. This is precisely how Esa wanted to it be, but not necessarily what they had planned for. As IXV programme manager Giorgio Tumino says, the craft had been engineered "beyond the worst case scenario you have, and in reality we found that we had as much as 100 per cent margins in some areas".
One special point of pride for Tumino was the performance of the guidance, navigation and control systems, which proved to be "accurate down to a couple of metres".
IXV is a big step for Esa, which is best known for daring one-way ticket robotic missions like Rosetta and Earth observation satellites, and had only dabbled in capsule technology beforehand. The agency is now firmly following its own re-entry roadmap, with plans to produce a successor vehicle with landing gear.
Would Esa be prepared to put astronauts in something like the IXV? "Based on the data, we'll consolidate our margins, there would be some changes required in architecture, you need redundant equipment, and we know how to do that," says Thales Alenia Space's IXV manager Roberto Angelini, who oversaw its production.
So? There's a long pause. "Why not?" he says.
Orion's belter of a flight
Across the Atlantic, Nasa's re-entry prototype Orion is also generating a buzz. The first mission of the programme last December sent the capsule up to 5,800km, 15 times higher than the ISS, and further than anything built to carry humans since Apollo 17 - a grand total of 43 years ago.
The flight also delivered a deluge of data, providing new impetus for the US space agency, which has been hankering for a re-entry vehicle since the Shuttle retirement in 2011. Orion doesn't have the instant showroom appeal of the IXV. It's a classic, Soyuz-style flat-bottomed capsule that isn't anywhere near as manoeuvrable as a winged vehicle or lifting body, but which tends to get the job done if you have a big patch of ocean or desert to aim for. The reason for that approach is simple: Orion has been designed with crew in mind. "Safety is an enormous driver in the decisions that we make," says Lara Kearney, deputy manager for the Orion Crew Module, based in Houston, Texas.
The heat shield on Orion uses a mildly updated version of a substance called Avcoat, developed for the Moon missions and their Apollo command module. Kearney describes it as being like a "gooey gum that is squeezed into 360,000 little honeycomb holes on the heat shield, to create this overall large surface that is then sanded down smooth". This ablative material is charred away during re-entry, protecting the capsule.
The Orion capsule is now being analysed at Marshall Space Flight Centre in Alabama. The test flight went according to plan, but there was sign of more charring in an area where the re-entry capsule separates from what's called the service module - the part of Orion that provides power and propulsion in space, but which is jettisoned before re-entry.
Orion aims to fly with astronauts on board in the 2020s. It may aim for the Moon, an asteroid, or even Mars, although Kearney stresses that "Orion will never land on Mars, it will support the crew on the way out and the way back".
Besides the plans of companies such as Mars One to send people to our red, rocky neighbour on one-way tickets, researchers are also assessing the possibility of Mars being our next big target for human exploration. "The people operating Curiosity [rover] on Mars can't wait for manned missions to set foot on the planet," says Esa rookie astronaut Thomas Pesquet, who heads to the ISS in 2016. "They say it's so frustrating, it's so slow, they want to get more. Of course human missions to Mars are complicated, but then the return is a 100 times better."
As ever, that ambition has to meet the reality of re-entry - this time, on Mars. Landing there is even more fast and furious than landing on Earth. 'Seven Minutes of Terror' was the catchphrase for the landing of Mars Science Laboratory (MSL) robotic probe that brought Curiosity rover to the Red Planet in 2012 - with a seven-minute descent and a 14-minute time delay, the Nasa engineers just had to cross their fingers and hope for the best. Luckily, it worked, and the rover landed undamaged. Some previous probes were less lucky. UK-made Beagle-2, a dustbin-lid-sized device, plonked onto the Martian surface in 2003 but never sent a signal home. It was eventually spotted in January this year after some very patient individuals sifted through a lot of pictures taken by a Nasa Mars orbiter.
If the challenge of landing on Mars is already huge for a tiny rover, it's even more so for a human crew.
The trouble is that Mars's thin, CO2-rich atmosphere is a "strange mix", says Nasa's Mars programme engineering manager Rob Manning. "There's too much atmosphere to land like we do on the Moon, where you take your vehicle and you fire the engine backwards until you hit the ground, and there's too little atmosphere to do it like we do on Earth," he says. The result is that engineers have to build hybrid spacecraft that look like Transformers, turning from cone-shaped Soyuz-style capsules to booster-wielding Moon landers.
In 2016, Esa will test its Mars landing setup in the first of the two ExoMars missions - a project to hunt for signs of life - preparing for 'six minutes of terror' in the process. In the time it would take you to make a cup of tea and log on to your computer, the lander will go through the distinct steps of a Martian landing. It'll first use the heat shield and the atmosphere to slow from the 12,000km/h velocity of the orbiter; then at 2,450km/h it will release a supersonic parachute and eject its heat shield to reveal a retro rocket; it will then fire at about 1,400m above the Martian surface at speeds of around 290km/h.
Finally, just two metres above the surface, the engine will switch off and ExoMars will freefall to the surface. A unique carbon-fibre and aluminium crashable structure is meant to cushion the impact and allow its instruments to 'phone' home.
The trouble is, "robots don't mind hitting Mars with the equivalent of 8 to 14 Gs of impact force, but that would be bad news for a human being," says Manning. A solution might be, he adds, to "fire rockets backwards while going supersonic, at four times the speed of sound". That's similar to what SpaceX has been doing this year when it tried to bring the first stage of its rocket back down to Earth and land it on a barge. So far they've managed to hit the barge, but not get the rocket to land upright. Until they get that right on Earth, it's going to be hard to persuade any astronauts to try similar high-speed stunts on neighbouring planets.
Nasa is also working on early-stage technology to get people and their heavy supplies to the Red Planet. One design is a supersonic-doughnut ring called the LDSD, or Low Density Supersonic Decelerator. It has inflatable sides on a very broad, flat capsule, giving the kind of surface area that would be needed to land on Mars. Manning is involved with that project too, and describes how during their first test last year the inflatable part worked perfectly at 55km altitude and 4,600km/h, but the supersonic parachute 'got ripped to shreds'. They're not giving up, though - another test of the LDSD is scheduled for June.
The process of cracking the business of re-entry, descent and landing continues. We can learn from the passengers, people like De Winne, who vividly describes the transition from "quietly just flying along" outside the atmosphere to "this screaming noise that you have from the air coming by" during re-entry. And we can take lessons from the test data, which proves that what you see in your computer model doesn't necessarily match reality. As Thales Alenia Space's Angelini put it: "What we learn from our job is different to the science-fiction movies."