Next Post Previous Post

25 Jul 2009

Composites’ famously low impact tolerance

Posted by Hans van der Zanden

The launch of the Space Shuttle Endeavour last Tuesday once again draws attention to the vulnerability of the heat shield. Video and photo images taken during the launch  revealed debris that broke loose from the external fuel tank. To the surprise of NASA engineers who commented “We don’t understand why that happened”, foam loss can be observed on seventeen different areas of the external tank.  Debris impact can cause serious damage to the Shuttle’s heat shield during the first 135 seconds after launch, when the Shuttle is still in the dense lower regions of the atmosphere. One debris impact occurred at one minute and 47 seconds into the flight and eroded the black outer coating of heat shield tiles in three areas. Another impact eight seconds later produced another area of outer-coating erosion. Preliminary inspections showed no obvious problems, but data analysis will take several days to complete. Impact damage is also a major problem with all composite aircraft, which are significantly at more risk than the Space Shuttle.

Launch images of the Shuttle Columbia in 2003 also revealed a serious impact during take-off. This was ignored, as was standard procedure since impact damage had occurred on all previous 112 missions. In some cases, damage was often severe, but was handled as just another turn-around issue that was dealt with when each orbiter was prepped for its next flight, and it was by sheer luck that an accident did not happen on those first 112 flights. But during the 113th Shuttle flight, which was flown by Columbia and her seven crewmembers, NASA’s ignorance – or, rather, NASA’s reluctance to better understand what they were dealing with regarding debris impacts – led to the loss of Columbia and her crew.  (Further details on this can be found and downloaded in the appendix on the right.)

That the heat shield was vulnerable to impact damage was very well known. “The heat shield of ceramic tiles and the reinforced carbon carbon panels were not designed to be damaged in any way for any reason. That’s why the orbiter isn’t allowed to fly through rain, stay outside when it hails, or risk having workers drop tools on it”. Columbia was the first Space Shuttle to fly back in 1981, and even on its first flight had sustained heavy damage. More than 300 tiles had to be replaced upon completion of its first mission, but this was no surprise to the engineers who later acknowledged to the Investigation Board that they had known in advance that the External Tank “was going to produce the debris shower that occurred”. It is therefore difficult to understand why NASA paid no attention to the impact behaviour of the heat shield materials when the Space Shuttle was designed, even more so when shuttles returned with heavily damaged heat shields. At the time of the Columbia accident in 2003, only a rudimentary test program had been performed in 1999, and that was not even completed. The carbon-carbon panels that eventually led to the Columbia accident were never examined for impact response properties.

After the Columbia accident the fleet was grounded and NASA started a “return to flight program” that would last for over 2 years. For the first time, the impact response of the heat shield materials was studied in detail. With no robust test methods available, the program relied heavily on impact modelling. At first glance such modelling should not be too complicated. Only single impact events have to be considered, and just four impact materials were involved; that is, lumps of foam and ice against the reinforced carbon-carbon panels and ceramic tiles. These materials are relatively simple and characterization proved not all that difficult to calculate. Broader conditions could also be fairly accurately specified; that is impact geometry, impact velocity, and angle of impact. But impact was not well understood and the modelling soon became a very complicated issue. It took the hi-powered computers at NASA more than 2.5 million hours of calculating time over six months to perform the complex analysis. No question the results were essential for understanding the cause of the accident and contributed in significant ways to our understanding of impacts that have never been researched in such detail before. Unfortunately, calculated results could not be properly validated with physical results because of a lack of proper test methods. In the end the modelling proved only the obvious; that is, that it allows for the study of material impact behaviour in some detail, but does not alter material behaviour. The first return-to-flight, by the Shuttle Discovery, again experienced multiple foam loss during launch on the 28th of July, 2005. “The large size of some of the foam loss caused concern because they were much larger than analysis had predicted was likely” , and subsequent flights provided a similar picture. The Space Shuttle is just designed incorrectly, as is further explained in the appendix at the right.

It is therefore of utmost importance that as much imagery as is possible is obtained during each launch. For detailed inspection in space the Orbiter is now provided with a robot-arm with sensors that provide a limited ability to inspect the tiles underneath the Shuttle shortly after it reaches orbit. Photo imagery of the acreage tiles across the bottom of the Orbiter is also taken by the crew of the International Space Station before docking. Unfortunately, repair in orbit remains difficult, if not impossible. “Despite comprehensive efforts to develop TPS repair materials and techniques, the state-of-the-art technology in this area has yielded modest technology to support the capability. As a result, continued efforts do not hold promise of significant capabilities beyond those in hand”. But the astronauts now have at least a chance and in case a repair is not possible another Space Shuttle can be launched for a rescue mission. This is contrary to conventional aircraft that have to cope with the force of gravity instantly when serious damage occurs.

Aircraft are much more susceptible to impact than a space shuttle through accidental collision with ground handling equipment, tool damage, de-icer impact, moisture and rain, sand storms, runway debris, engine debris, blade loss and rotor burst, hail stones, lightning strike shock waves, bird strikes, meteorites, hard landings, busted tire debris, and wheel threats. These impact occurrences do not present much of a problem with aluminium aircraft, but pose a serious safety risk with aircraft where the skin is made completely out of composites that are much more vulnerable to impact than the ceramic tiles and the carbon-carbon panels of the Space Shuttle. Composites can even get damaged with low velocity impacts, for example by dropping a tool or by walking on a wing. To put it bluntly, with all-composite aircraft the window glazing provides much better impact response than the composite skin itself, as is discussed in some detail in “An impossible dream”. This means that all-composite aircraft will experience a lot of impact damage during their service lifetime; damage that, because of the intricacies of the composite material, accumulates and is compounded when repaired. It appears that this has been largely ignored by Boeing and Airbus in a way that shows striking resemblance to the Space Shuttle. One concern is multiple delamination that might interact and link up to form larger delaminated areas that can hardly be detectable.  This is essentially similar to widespread fatigue damage that can occur with aluminium aircraft, were cracks link up to form a larger crack that can no longer be contained, as happened during Aloha flight 243, a Boeing 737 that lost a large part of its upper fuselage during flight. It has to be awaited for the results of the investigation to understand what caused a large hole to appear in the 737 fuselage of Southwest flight 387 last week.

Another concern that has not been studied in any detail is the effect of multiple impacts, that is a sequence of repetitive impacts at locations very close to each other. An impact face that moves at high velocity through a hail storm can experience such repetitive impacts. It is also not clear how two or more impacts that occur simultaneously close to each other affect one another, but could intensify the delamination process. Impact geometry has not been studied in great detail, for example, the effect of rounded and sharp impact, and self-rotation of the impactor. Other influences include the stressed or vibrating impact face or already damaged impact areas, and of course temperature and humidity, to mention only the most important and the most obvious. This makes clear that modelling can only provide an approximation of what happens with real world impacts.

Most severe are hail impacts. Hail stones can be more than 4 inch in diameter and impacts are continuous, repetitive, and widespread, and are likely to occur at cruise speed; that is, some 800 km/hr or 497 m/hr. Such impact sequences can cause severe damage, even with aluminium aircraft as the following picture shows. The question becomes how a composite nose will behave under such circumstances. Given composites’ low impact tolerance, the nose and other critical sections of all-composite aircraft should be tested for such impact events in a way similar to how the engines are tested. These are run at full speed when a load of ice chunks are released in front of the engine, as is illustrated in the following video. The problem is that the hail is not at cruise velocity. Additional tests are performed to simulate this effect, but it involves again a single impact event. The vanes are constructed out of composite but are provided with titanium edges that encounter the impact. The vanes are running at such high speeds that any particle is completely grinded and accelerated upon contact with these titanium edges; that is, before the incoming material reaches the composite faces that are only loaded by centrifugal sliding (and some impact) of very fine particles. The composite vanes would be severely damaged if they were not rotating.

Unfortunately there is no test method available yet that can throw a continuous spread and stream of hail stones at high velocity towards an engine or a nose or wing section, as is further explained in “Sudden impact”, which can be downloaded at the right. Physical testing involves only single hail-stone impacts with the aid of a pressure gun, a rather complicated test method. Also mathematical modelling is limited to such single impacts and of little value, as has been explained before. Real world testing is, however, most important as the tests that were performed after the Columbia accident showed. Analyses indicated that the heat shield was damaged through impact of a rather large piece of foam against a carbon-carbon panel. For many engineers at NASA this was difficult to believe. So, a test was performed when the impact of piece of foam against a carbon-carbon panel was accurately simulated in a 1:1 test. “I don’t think anyone expected to see a 16-inch square hole”, one of the engineers reported later. “In the blink of an eye, there it was, and hundreds of people immediately came to terms with how much damage a piece of foam can do”. The Columbia accident is further described in the appendix that can be downloaded at the right.

It is therefore of utmost importance that an accurate and robust method of testing becomes available for 1:1 impact testing of all composite aircraft, or at least of full scale parts that are susceptible to impact. This author is working on a simple test method that makes it possible to simulate such real life impact sequences, 1:1 in a fully deterministic way including real environment conditions, that is temperature of minus 60 degrees Celsius. This method is much more simple than the one described in “Sudden impact” and will be presented on this site soon. It will then be possible to see the real effects of hail impacts and compare impact response of aluminium and all-composite structures. That is, when Boeing and Airbus are prepared to perform such tests. It will be most interesting to see to what extend the fibres offer protection.

Leave a Reply