In previous posts about Mars and the difficulties of landing on other planets, I’ve mentioned that the red planet is a veritable graveyard for unmanned spacecraft – two-thirds of all Mars-bound missions have failed. In some cases, engineers are able to figure out why. The key is telemetry and a thorough understanding of the spacecraft. With the right tools, the failures act as lessons, illuminating unknown challenges so mistakes aren’t repeated. But the data doesn’t always come back, leaving engineers with little to go on if a mission fails. The worst-case scenario, and not an altogether uncommon occurrence, is the total loss of communication with a spacecraft with no indication of what went wrong. (Left, an artist’s rendition on the ESA’s Beagle 2 as it enters the Martian Atmosphere.)
Two missions in the last 12 years have failed rather spectacularly: NASA’s Mars Polar Lander (MPL) and the European Space Agency’s (ESA) Beagle 2. Their failures were in very different circumstances, however, and a comparison of events illustrates the importance of pre-launch testing and constant telemetry for the success of a mission. These two cases also demonstrate the amount of work needed to successfully land a spacecraft, especially when the target is millions of miles away.
The first missions to Mars were part of the space race; NASA’s Mariner program flew by the planet in the 1960s before the Viking program sent two stationary landers in 1976 (right). But when Viking failed to find obvious signs of life, Mars fell out of favour as a destination for unmanned flight.
The bulk of work done on Mars throughout the 1980s was in labs on Earth, testing and reviewing the data collected from Viking 1 and 2. Missions to the red planet resumed in the early 1990s with the aim of putting a satellite in orbit around the planet. After a failed first attempt with the Mars Observer in 1992, the Mars Global Surveyor successfully entered orbit around Mars in 1997.
Interest in Mars was reignited in August of 1996 when scientists at NASA made a striking discovery – a meteorite found in Antarctica, determined to be of Martian origin, contained traces of fossilized bacteria. Suddenly Mars wasn’t just a long-dead planet, it was a long-dead planet filled with vestiges of life buried beneath its surface.
In the wake of this discovery, two missions were born: from NASA, the Mars Polar Lander, and from the ESA, the Beagle 2 lander.
The Mars Polar Lander (left) was proposed in 1994 as part of the Mars Surveyor Program that aimed to send a spacecraft to Mars every 26 months for a decade – MPL was set to be the first landing of the program in December 1999. Its goal wasn’t to find life but rather to understand the climate history of Mars. Its landing point was an area suspected to have at one time been wet – water is, after all, one of the necessary building blocks for life. The mission also supported what is still NASA’s long-term plan of a sample return mission by identifying the types of samples worth bringing home.
MPL used the same retrorocket-powered landing system as the Viking landers. During its interplanetary journey, the MPL would be protected in a conical casing, the base of which was the heat shield that would protect the lander from friction as it began its entry through the atmosphere. 5.4 miles above the surface, the main parachute would deploy, slowing the spacecraft from 0.3 to 0.1 miles per second. With the descent slowed, the heat shield would jettison, exposing the lander and the landing radar.
0.8 miles above the surface, the lander would separate from the backshell – the upper part of the casing that holds the parachute. Firing its retrorockets located on its underside, the lander would slow further to a descent rate of 0.05 miles per hour. At this point in the landing sequence, the lander’s legs would deploy. Just before touchdown, the rate of descent would slow almost to a standstill – 0.001 miles per second. As soon as the sensors in the lander’s footpads detect contact with the surface, the retrorockets would shut off. MPL would be set to begin its stay on the surface. (Pictured is an artist’s conception of the final descent stage of the Mars Phoenix Lander that used the same retrorocket design to touch down on the surface.)
NASA did routine checks on the spacecraft during its interplanetary journey, and as the MPL neared Mars it was in good health. Just prior to atmospheric entry, the spacecraft reoriented itself flawlessly, beginning the expected loss of communication with the lander – its antenna would face away from Earth throughout EDL. Contact from the lander was expected about 45 minutes later when it was safely on the surface with its antenna reoriented towards earth.
But the expected transmission never came. NASA attempted to contact the lander for weeks, but to no avail. Although later contact efforts were made, MPL was declared lost on January 17, 2000.
The orbiting Mars Global Surveyor began the hunt for the lost lander not long after the loss of communication by taking images of the intended landing site. Even the highest resolution images available, however, produced no clear evidence of the lander nor shed any light on its fate. (Left, the first image taken by the MGS satellite of the Mars Polar Lander’s landing site.)
And so an investigation was set up at NASA’s Jet Propulsion Laboratory (JPL), the team who actually built the lander, to investigate possible fates of the MPL.
One problem was immediately evident – the lack of telemetry gathered during EDL left NASA in the dark. The lack of telemetry was a conscious decision. In an effort to keep costs down, anything deemed superfluous to the mission was cut; the directive from NASA was that “no resources would be expended on efforts that did not directly contribute to landing safely on the surface of Mars.” This meant EDL telemetry, although vital for engineers’ understanding of the critical phase but inconsequential for the spacecraft’s actual landing, was unnecessary.
Luckily, NASA had done extensive testing of each piece of the MPL and could reverse engineer the problem – use their knowledge of the systems to determine possible failures. The investigation was broken into three principle areas. First, system fault analysis or a consideration of the most significant risks of the mission; second, fault tolerance or the ability of the system to cope with adverse circumstances; and third, margin characterization or an understanding of the margin of error between the systems performance and the point at which it fails.
Using these three factors, the investigation team was able to come up with a number of possible reasons the lander failed. Uneven surface features could have affected the radar’s readings, causing the lander to touch down on an uneven or dangerous surface. The heat shield could have failed because of a manufacturing defect, an impact with a micrometeoroid, or simply inadequate design parameters. A design flaw could have caused a loss of control and stability in the lander in the terminal stages of its descent. An imbalance between thrust levels in the retrorockets resulting in an off-centre mass could have brought the lander down in an unrecoverable position such as on its side. The landing site might have been unsurvivable; it could have landed on a precipice and rolled into an unknown chasm. The backshell could have landed on top of the lander and crushed it. Plausible as these explanations may be, they were inconsistent with test data from the MPL’s development.
One possible fault, however, was supported by MPL test data; the investigation determined that the most probable cause of failure was a fault in the landing sensor. Each of the lander’s legs relied on a magnetic sensor to determine when the lander made contact with the Martian surface, which would in turn trigger shutdown of the retrorockets. Each sensor reacts to a hyperextension of the leg’s joint as the weight of the lander settles onto it. But there was a flaw in the sensors’ design.
During the interplanetary journey, the MPL’s legs were stored folded. The force of deployment caused the same hyperextension of each leg’s joint as landing. Data pulled from the MPL development and deployment tests indicated that a false touchdown registers with the lander’s onboard systems when the legs are deployed, tripping the sensors and shutting off the retrorockets. The software couldn’t distinguish between the landing legs deployment and a touchdown. (Left, the Mars Phoenix Lander’s legs are seen folded underneath its body for the interplanetary journey. While this isn’t the same as the Mars Polar Lander, it gives an idea of the legs’ configuration prior to deployment.)
The review board painted a picture of the likeliest course of events: The legs deployed 0.02 miles above the surface when the lander was descending at a rate of 0.008 miles per second. The fall from that height would have accelerated the lander to 0.0.1 miles per second at the time of impact – much faster than the ideal landing speed of 0.001 miles per second. This would explain the lander’s good health going into EDL, and the total lack of communication after touchdown. It was also a preventable problem. A software program could distinguish between, say, between a brief touchdown signal from a snapping joint or a prolonged signal from the settling lander. Unfortunately, no one had included such a program into the lander’s systems.
While the MPL was lost, the mission wasn’t a complete failure – determining the cause of death of the lander gave NASA an indication of how to not repeat the same mistake. The MPL mission was revived as the Mars Phoenix Lander in the mid 2000s; Phoenix landed successfully near subsurface water close to the Martian north pole in 2008. (Right, the Mars Phoenix Lander launches on a Delta II rocket.)
Other missions, however, experienced failures that didn’t leave any traces of fault. One such mission was the ESA’s Beagle 2 lander, part of the Mars Express program announced in 1997. Set for launch in 2003, the lander was designed to look for life through geochemical and atmospheric analyses. The spacecraft was named after The Beagle, the ship that took Darwin through the Galapagos Islands. Within the name was the grandiose promise that the ESA would do for space science what Darwin did for evolution.
Beagle 2 (left) was designed to land at an area on Mars called Isidis Planitia – a large, flat sedimentary basin – on Christmas day 2003. The choice of landing site was deliberate. It sits at a low altitude where the depth of atmosphere above the surface is greatest. This allowed the lander to take advantage of the natural braking from atmospheric drag as well as a longer descent with the parachute. Planning to use the atmosphere to its advantage, the ESA developed a daring EDL profile.
Beagle 2 would make its interplanetary journey in a protective case, the base of which is the heat shield that protects the lander as it enters the Martian atmosphere. After moving though the upper atmosphere, a drogue chute would deploy to begin slowing the lander’s descent. At the same time, lander forcibly separates from the protective case and jettisons the heat shield. Once free form the casing, the main chute deploys from Beagle 2 itself.
0.1 miles above the surface, a series of airbags inflate around the lander forming a protective cocoon. Upon contact with the surface, the parachute jettisons and the lander in airbags is free to bounce and roll along the surface until it naturally stops. Irrespective of its landing position, a hinge opens the clamshell-like cover around the lander, automatically righting it and exposing it to the Martian environment.
The method is simple but risky – there is nothing the control the descent of the lander but a single parachute, and nothing to protect it save the airbags. Arguably, the decision to pursue this landing profile is where Beagle 2’s problems began. Adding to the problems was the alarming rate at which the project progressed; the ESA had adopted the “better, faster, cheaper” attitude reminiscent of NASA in the 1980s and rushed their development.
From inception to delivery, the parachute for Beagle 2 was developed in 15 weeks. The chute was built by Astrium to meet the qualifications and standards of Martian flight models.
A second chute was built and tested in the Arizona desert, the environment most consistent with conditions on Mars. The first drop test was successful, as were further drops over the following two weeks. Equally successful was a demonstration of the parachute’s rapid extraction from its pack at speeds of up to 90 miles per hour. A final test of the parachute’s strength was done with the simple but severe method of towing it behind a truck.
The airbags used to cushion the lander’s actual impact and travel along the surface were tested once – the configuration was dropped straight down onto a flat surface.
These test were done flawlessly, but they hardly put the parachute and landing bags through the rigours they would likely encounter on Mars. Part of the problem is the impossibility of simulating a Martian landing – a fascinating problem I will tackle in a later post.
But the more striking problem is the ESA’s limited testing. All drop tests were done in two weeks. The extraction and tow tests, as well as the airbag test, were each done only once. As a point of comparison, when NASA tested the airbags that delivered Spirit and Opportunity to Mars, the airbags were dropped hundreds of times at different angles over flat surfaces and on rocks. Only when no test revealed any flaws was the system deemed flight ready. And the system performed beautifully during the mission. (Left, a multiple expose shot of an airbag drop test for the 1997 Mars Pathfinder mission. 1995.)
The same, unfortunately, cannot be said for Beagle 2. The last contact ESA had with the lander was on December 19 when it separated from the Mars Express transport craft. At the time, the lander was in good shape and all the systems required for EDL were well within limits for a safe landing.
Then it went silent.
The lander didn’t actually communicate directly with Earth. Instead, it went through third party spacecraft – the primary mode of communication was via the Mars Express spacecraft. Secondary was communication via the Mars Odyssey orbiter, and there was the slight chance that Jodrell Bank Observatory near Manchester might be able to pick up a weak signal from the spacecraft. Neither the orbiting spacecraft nor the observatory turned up any information on Beagle 2. Orbital imaging of the planned landing zone hasn’t yet revealed the fate of the lander, and with such limited testing, there are too many possible reasons the mission failed that engineers weren’t able to nail down a single cause of the failure.
The fates of NASA’s Mars Polar Lander and the ESA’s Beagle 2 illustrate not only the importance rigorous testing and an intimate knowledge of the spacecraft, but having a means of communicating with the spacecraft. NASA’s next mission to Mars, the Mars Science Laboratory (MSL), has an even more daring EDL planned. With future Martian exploration hinging on the success of this mission – another story best left told on its own – the worst case scenario is for the spacecraft to disappear without any indication of its failure. MSL is set to launch in November 2011 reaching Mars in the summer of 2012. There’s nothing to do but wait and see what happens.
1. Fredric W. Taylor. The Scientific Exploration of Mars. Cambridge. 2010.
2. Steve Squyres. Roving Mars. Hyperion. 2006.
3. Andrew Mishkin. Sojourner: An Insider’s View of the Mars Pathfinder Mission. Berkley Trade. 2004.
4. ESA’s site about Beagle 2.
5. “NASA JPL Mars Phoenix Lander HD Animation”: http://www.youtube.com/watch?v=KHAsRjcLP4o [Accessed March 25, 2011]
6. “1998 Mars Missions. Press Kit, December 1998″ NASA.
7. “Report on the Loss of the Mars Polar Lander and Deep Space 2 Missions; JPL Special Review Board” JPL. 2002.