Conclusions
11.1 Introduction: causes of product failure
The cases discussed in previous chapters have many common components, apart from the obvious point that they all involve failed polymer products of one type or another. The various product failures were caused by defects which arose in several different ways, including:
But many of the failures arose through a combination of such problems, where one kind of defect such as cold moulding was exacerbated by another, such as design or geometric stress concentrations.
Polymeric materials exhibit a spectrum of physical and chemical properties quite different from conventional materials such as metals, ceramics or glasses, low density perhaps being among the most prominent. It makes polymers good candidates for any product which involves transportation, such as vehicles of all kinds. Ease of processing to shape is a second feature which makes them attractive for replacement of existing components. Owing to their relatively low thermal transitions and their high viscosities, they can be moulded into very complex shapes so that many parts can be replaced by a single part, thus saving substantial manufacturing costs. On the other hand, they are, as a group of materials, flammable and subject to degradation, especially by oxidation from a range of active agents.
11.2 Poor manufacturing methods
While part consolidation can be beneficial in simplifying product designs, there are many caveats. Polymer must flow evenly and smoothly to all parts of the tool cavity during injection moulding, and must be hot enough to prevent the formation of weld lines. Although not always detrimental, when they occur in a zone that is highly stressed in service, they become defects. The entire product may fail when this happens, so great care is needed in eliminating or moving them from critical areas of a moulding. The case of the oxygen bottle security caps in Chapter 8 is a case in point: widespread failures occurred in storage simply because a weld line moved to the critical part. The device functioned by the cap bending about the ‘living hinge’ when initially attached to a bottle, and it then had to remain in place until the gas was needed at the hospital. If the hinge failed during storage before even being applied, it had failed its main purpose. The problem produced a crisis at the company as thousands of caps failed during storage, and had to be scrapped. A systematic approach to the problem was needed rather than a knee-jerk reaction.
11.2.1 Faulty moulding
The fault lay in control of the moulding machines used to make the caps in a single step. There were three gates to the tool, and if equal flow prevailed then the weld line ended up in the hinge. Changes in barrel or tool temperature had clearly changed the flow patterns within the tool, and had to be monitored more closely and accurately to solve the problem. The investigation also produced a simple quality control test to determine the strength of the device after fitment using simple ergonomic ideas. On discovering the problem, the moulding company jumped to the conclusion that the material itself was at fault, but independent analysis showed little difference between the alternative supplies, and pointed to a quite different cause.
The polymer concerned in the example was polypropylene, a common material used widely in many different products. With other polymers, however, other types of problem can occur when tool temperatures are below recommended levels. Polycarbonate is just such a material where cold tools can freeze the molecular chains into non-equilibrium conformations so that it makes the product sensitive to brittle cracking when exposed to organic solvents (and many non-solvents).
11.2.2 Assembly problem
The problem was encountered by widespread failures of miners’ lamp battery cases underground when in use at the coalface, so endangering the safety of individual workers. At one point in time, the problem was so serious as to bring an entire colliery to a halt for lack of safe lamps for the face workers to use. Against the advice from the material manufacturers, the sub-contractors moulded the cases and tops into cool or cold tools, so producing very high levels of residual strain. When it came to join top to base after insertion of the contents, brittle cracking was encountered. Unfortunately, reject levels on the production line were relatively low, so the true extent of the problem was not fully appreciated. Then cracked lamps were found at the collieries, and colliers themselves suffered from failed lights and spilt acid.
Cracks were growing slowly through the polymer and lamps passed as fit for purpose after manufacture ultimately failed as the cracks grew through the walls and led to acid leaks. The cracks were initiated when the solvent used to weld the parts together interacted with the polymer to create the cracks.
The root cause was traced by examining the transparent parts using polarized light. It revealed that the plastic sheet showed high levels of birefringence caused by chain orientation. It in turn was due to the moulding conditions used to make the components: cold tools effectively quenched the molten polymer when it entered the cavity. When that highly oriented polymer met solvent, cracks were created and grew with time.
The solution lay in changing moulding conditions to use hot tools, a move involving some expense because the tools needed modifying to accept hot oil as the circulating fluid. From then on, the failures started to drop away, although other problems were found in the design of the cases.
11.2.3 Medical devices
Polycarbonate has grown in its applications, one being in medical devices such as catheter connections. Much new technology has been introduced into hospitals in recent years, and many devices have failed at critical times when being used with patients. A recent example involved moulded connections between catheters, devices designed to allow staff to feed different fluids to poorly patients such as premature babies. The problem of brittle cracking was discovered by nurses when fluids began to escape but there was a more serious problem of bacterial contamination of the lines through penetrating hairline cracks. One such baby became infected after numerous connectors were replaced. Fortunately, the mother had retained one such set of connectors, and examination showed them to be cracked, the cracks penetrating through the wall, although not, apparently, into the line itself. The polycarbonate of the connectors had probably been moulded into cool tools, and when common hospital liquids contacted the outer surfaces, cracks had been initiated. Similar failures had been widespread through the UK and the manufacturer had not investigated the incidents very thoroughly. They had also destroyed the failed samples, so denying other investigators the opportunity.
So despite the existing knowledge concerning the problems of moulding polycarbonate, the manufacturer was blissfully unaware of the perils of that course of action, and put patients at risk of infection. The baby who became infected suffered brain damage, but was compensated by a large award. Unfortunately, brain damage cannot be reversed, and the young adult will need ongoing assistance.
11.3 Poor design
Both products also showed basic flaws in their design. The battery cases possessed many sharp external corners as well as sharp inner corners, all of which weakened the final product substantially. Most tough polymers exhibit the same problem, known as notch sensitivity: the sharper a notch, the lower the strength of the product. When lamps are dropped, it is the inner lower corners of the boxes which act as notches, and from which brittle cracks will propagate. Sharp corners occur in numerous polymer products, and battery cases are just one example of a practice that is widespread, as many of the cases in previous chapters have already shown. They include:
• Chapter 3: corner at shoulder of breast tissue expander
• Chapter 5: screw thread tips on ebonite handles, corners on truck and miners’ lamps battery cases
• Chapter 6: shoulder of PVC rising main, screw threads of acetal fitting
• Chapter 8: corners on angle grinders, and stepladders
• Chapter 10: screw threads on busbar casing, serrations on bike carrier shell, babycatch.
There is another danger with sharp corners: when exposed to liquids, it is where brittle cracks caused by ESC or SCC will often start. Even when the liquid is removed, a meniscus tends to remain at such corners, so increasing the chances of cracking. When exposed to gases like ozone, attack will tend to start at corners simply because any applied stress will be greatest there. Just this was found in the fractured NBR seals from the semi-conductor factory in Japan.
It is inexcusable to leave sharp corners in mouldings, and engineering drawings should specify a minimum radius of curvature. However, more often than not, such recommendations are absent, and toolmakers leave external corners on cores very sharp. It is surely good practice to smooth such corners before moulding products, and so strengthen the final component. It is also one of the simplest and cheapest tasks to perform in the toolroom.
But holes also represent stress concentrations, and they are often inevitable for making attachments or connections. It is entirely predictable that such holes will be stressed in service if the connection is live, such as the bucket lugs of the failed bucket case of Chapter 5. In this case, the new approach adopted incorporated the lugs with the wall below, so extra support was gained. The new design was expensive, owing to the new tool needed, but produced a much safer product and it has been adopted by all manufacturers of such products. The exposed lugs were at the end of the flow path, so weld lines were likely to be formed there, and so weaken the bucket yet further. Other products where cracks were started at holes included the broken catheter tip from Chapter 3, but here the material itself had become embrittled and the catheter failed where the local load was greatest at the edge of one of the bleed holes.
Even scratches can prove the undoing of a product. The WC handle discussed in Chapter 8 failed from scratches made by the user on its chromium-plated front. The scratches allowed strong cleaning agents such as bleach to attack the underlying ABS plastic, as well as acting to concentrate the stress locally. A brittle crack grew slowly at each use of the handle. It finally fractured when the crack was too large for the remaining solid material to support the applied load.
11.3.1 Stress concentrations
The way in which load is distributed in a component is vital to its performance, and when the load is concentrated at a critical zone, then problems can be expected. Good design during product development should examine the probable load paths by mechanical testing to find those zones and then change the geometry of the design to ameliorate and lessen the stresses there. Often the exercise is very simple, as in the example of the battery cases, where smoothing or radiusing the corners reduces the danger of brittle cracking and so strengthens the final product. More complex component geometries can be tackled either empirically or by using compilations of stress concentration diagrams or relevant equations (1,2). Finite element analysis can also be useful during prototyping for predicting critical areas of a product (3). The theory underlying stress concentrations is of course not specific to any material, but there are extra precautions needed when designing with polymers.
For example, moulded products show an extra level of complexity: it concerns the way flow occurs within the tool cavity. Holes are a particular problem because the flow has to split as it encounters the edge, and then recombine at the further edge. If the polymer melt is at too low a temperature, a weld line forms. If the temperature of the melt is too low then the weld line represents a proto-crack from which a real crack can start when the product is loaded.
One important feature of flow analysis (where finite element analysis can also be applied in tools) is its use in designing tools. Detailed information is needed on the grade or molecular weight of polymer being considered, since it will determine the melt viscosity and hence the flow rates into the tool at various temperatures. The presence of sharp corners can affect melt flow by constraining the flow. A good example of the problem is the occurrence of high degrees of chain orientation at such obstacles. So the severity of a geometric stress concentration may be increased because oriented polymer will tend to fail earlier than unoriented material. The effect was shown in Fig. 5.41, where a highly birefringent zone occurred next to a belt loop corner.
Such stress concentrations are the weakest points in designs, and from which cracks are likely to form in adverse conditions. Such conditions include:
• overload under a steady stress
• stress corrosion, or exposure to an aggressive chemical
• environmental corrosion, or exposure to specific organic fluids.
Elimination or amelioration of stress raising features are therefore the sine qua non of the designer, especially when using polymeric materials.
11.4 Poor choice of materials
Like many other materials, polymers have strictly limited temperature stability owing to glass transition temperatures or melting points, where physical properties like stiffness and strength diminish greatly. The major difference from other materials is the much lower temperatures at which these transitions occur, typically from about 100 °C to 250 °C for most common polymers. Even at lower temperatures, many polymers will oxidize fairly rapidly, so degradation can be expected, a process that results in lower molecular weight and so lower strength. Polymers must therefore be selected knowing the maximum temperature in which they will operate.
One of the most striking cases involving inappropriate choice of a polymer concerned the composite tank discussed in Chapter 4. The tank was designed to store 100 tonnes of water effluent at a temperature in the nineties centigrade, but was built from polyester and chopped glass fibre. The polyester showed a Tg of about 70 °C, so it was unsuitable for its purpose. It failed catastrophically about a year after installation. In addition, any distortion of the walls was hidden by an integral bund, so there was no forewarning of the impending failure.
But other inappropriate polymers have been used in products where the results of failure can be damaging to the user, such as in knife handles. The example of small cutting knives discussed in Chapter 8 showed that the polymer used in one of them, apparently HIPS, actually showed all the characteristics of polystyrene, a brittle polymer unsuited to highly stressed products. When it failed near the tip, the user suffered a severe cut to his finger. Another example used a tough material, polypropylene, but it could itself be cut easily and another accident occurred. The design could have been improved very easily by incorporating a metal insert to prevent the blade slipping.
An example of a component which need not have been used at all was the luggage trolley of Chapter 10. A polypropylene moulding was stapled to the hollow steel frame to locate the bungee cord used to secure the luggage. In two cases, it fractured suddenly under load from the cord, causing severe personal injuries to the face and eye. But the device was entirely superfluous since the cord could simply have been knotted around the frame, as indeed it was after these accidents had occurred.
The problem can occur with any material whose role is critical, as the example of the poorly selected sealants used in the ducting of the fire brigade training centre in London. Many of the sealants absorbed hydrocarbon oil used to create smoke in the building, became plasticized and semi-liquid. They ran downwards and so broke the seal in the ducting, allowing the smoke to soak into insulation, and a real fire was created. The sealants should have been tested beforehand, and the unsuitable ones rejected for use in this application.
11.5 Environmental stresses
A related category of product failure occurs when materials are affected deleteriously by their environment, a problem which is widespread among all the different types of material available to designers. Corrosion is a very serious problem with steel and many metals, the attack frequently occurring insidiously at inaccessible points where water collects. If the component is loaded heavily, then stress corrosion cracking can occur, with sudden and dramatic results. The fall of the Silver Bridge in West Virginia killed 46 drivers and passengers on cars on the bridge at the time when it fell at midday on 14 December 1967 (4). The suspension bridge had been under-designed with only two tie bars in each link of the main chains carrying the roadway, so when one cracked, there was nothing to prevent the entire bridge collapsing. It fell completely in less than a minute. The root cause of the cracked tie bar took the form of a 3 mm long crack which had grown over the 40-year life of the structure when it became critical. It had grown from the lower bearing surface of a joint near the top of the chain where water had collected, and grew under the influence of residual stress left in the bar after manufacture. When rust formed in the crack, the expansion exacerbated the stress on the crack tip.
Polymers too can suffer from stress corrosion cracking (SCC), and are also sensitive to environmental cracking (ESC), a problem unknown in other material types. SCC is caused by chemical attack on the polymer chains, while ESC is essentially a physical effect where organic solvents penetrate the bulk polymer and initiate brittle cracks (5). The fact that polymers can be degraded by a variety of chemicals is of course well known, although the details often remain obscure.
11.5.1 Stress corrosion cracking
There are several distinct kinds of SCC depending on both the polymer and the reagent which attacks the material. One of the most common forms of attack occurs when strong acids or alkalis cause hydrolysis of main chains, as exemplified by the cracking of the diesel fuel return pipe of Chapter 9. It was the cause of a road traffic accident and the cracked pipe was initially thought to have been the result of a knife cut by a vandal. But the pipe was buried deep in the engine compartment and difficult to access. The solution to the mystery lay elsewhere.
It needed only a small leak of sulphuric acid from a car battery to initiate a brittle crack in the nylon 66 connector, a crack which grew slowly as the line vibrated during use of the vehicle. It took about a week for the fuel line to fracture completely, and the diesel spewed into the road and caused several accidents. The worst of them severely injured a following driver when her car skidded on the diesel patch and careered into the path of an oncoming lorry. The sequence of events was determined by detailed examination of the fracture surface using ESEM, and the theory of SCC confirmed by analyzing the surface for the traces of sulphur left by the acid attack.
All so-called condensation polymers (alternatively called step-growth polymers) are made in the first place by linking monomers together via acid and base units, and it is these ester or amide units that are susceptible to hydrolysis in solid products. But there is a wide variation in their stability to acids and base liquids. For example, polycarbonate is stable to concentrated sulphuric and other acids, but highly sensitive to any type of base. Even very weak bases such as sodium carbonate solution will attack the material. Very strong alkalis such as caustic soda, attack the material very rapidly. By contrast, PET is resistant to alkalis but sensitive to acid attack.
Copolymers which possess polyester or polyether units, such as Hytrel, are also sensitive to hydrolysis, as the failure of elastomeric washers at numerous locations showed. They were attacked just by water, although the high temperatures of the radiators was decisive in initiating cracks. Those same temperatures also led to crystallization of the material, and so putting further strain on the components. It was a problem that should have been checked by testing the parts before adoption, and there were warnings from the material supplier (DuPont) of the drop in stability at high temperatures. The same problem can occur during moulding at the inevitably high temperatures of the molten polymer in the process machinery, with most step-growth polymers at risk unless dried thoroughly before processing.
Since hydrolysis involves chain splitting or scission, the molecular weight of the material falls fast, and since mechanical strength is dependent on molecular weight, the strength drops rapidly, too. It can be highly localized at the crack tip, so crack growth is enhanced; surrounding material may remain entirely unaffected, a frequent characteristic of brittle failure in many materials. As with all types of corrosive attack, crack growth is increased by any residual stresses or strains present in the product, often as a direct result of manufacturing methods.
11.5.2 Oxidation and ozonolysis
Most polymers of whatever type are susceptible to oxidation from a variety of powerful oxidants. Oxygen is omniscient and oxidation increases as the temperature rises, so polymers exposed to temperatures greater than ambient may be subject to attack. Polymers with secondary or tertiary carbon-hydrogen bonds are especially vulnerable because the free radicals formed at such bonds are more stable, and hence oxidation is favoured there. That means polymers such as polypropylene and polybutylene can be oxidized relatively easily, and anti-oxidants are added routinely at a very early stage in their manufacture. Such speciality chemicals do have a limited life, however. They act by absorbing the damaging free radicals, and the protection depends on the concentration of anti-oxidant present. As the molecules are depleted by reaction, the protection they provide falls, and there may come a time when there is no further product protection whatsoever. That was shown by the very widespread failures of polybutylene pipe in the USA, although they tended to fail after the acetal fittings used to connect the pipes together.
Ultra-violet radiation in sunlight can have a similar effect by interacting with the polymer chains to form free radicals, which then react further, usually resulting in chain cleavage and cracking. Even polyethylene, which is otherwise more resistant to oxidation than polypropylene, can suffer and so UV absorbing chemicals must also be added where continuous exposure to sunlight is expected. The failure of road cones and mancabs of Chapter 4 were examples of rotationally moulded products without that protection, and which failed very quickly when used externally, as was intended. A more complex example was given in Chapter 5 of storage batteries used in fork lift vehicles which had been hot welded, but where failure occurred because the oxidized tops were then exposed to strong sunlight during one of the Israeli wars. Another case involving a fractured catheter in Chapter 4 showed that the catheter had been affected by UV exposure at an early point in its history, and resulted in premature failure when in use in the hospital.
Chlorine even in very dilute form in water is a potent oxidizing agent, which is why it is so efficient in killing harmful bacteria. However, many polymers are equally susceptible to attack by the chemical (as are many metals). The flood caused by a broken acetal fitting (Chapter 6) was traced to chlorine attack after information from the USA revealed expert reports from a class action in Texas. The failure at Loughborough University caused extensive damage to computers because it happened at a weekend, so went undetected for some time. The fracture was old, but showed intermittent crack growth in the threads. The cracks had started from weld lines in the sample, and the failed moulding was probably a maverick because it could not be matched with current production from the mouldings. ESEM/ EDX later confirmed the presence of bound chlorine in the fracture surfaces. The US reports showed that acetal fittings had failed first in new plumbing systems installed in houses across the entire country. The parts failed in water supplies with chlorine levels as low as 1 ppm, but had been worst where water temperatures were highest.
Polymers possessing double bonds within their main chains are also liable to oxidation because the bond stabilizes the free radical formed when hydrogen is abstracted from the adjacent carbon atom:
The activated carbon atom can then react further with oxygen to produce a carbonyl group:
—CH = CH—CH*— + O2 → —CH = CH—CO— + O*
The carbonyl group can be detected using FTIR spectroscopy, a major tool for investigating polymer degradation, as previous cases have demonstrated.
But double bonded polymers, especially elastomers which form the majority of the type, are susceptible to another and more aggressive allo-trope of oxygen, namely ozone. The gas is formed near electrical equipment from sparking as well as silent discharge of current, but usually at very low concentrations (in ppb). But even such low levels of ozone will attack double bonds very selectively. They split the chain where a double bond occurs, forming carbonyl groups at the new chain ends. Cracks are formed, but can only grow under an applied load, as several cases showed, including failed NBR fuel pipes on Fiat cars and aircraft haulage vehicles, and NBR seals on semi-conductor fabrication units using pneumatic systems. Simply bending the vehicle fuel pipes was enough to promote crack growth, while the diaphragms in the pneumatic seals were flexed by air pressure when the equipment was working. However, preferential attack occurred at sharp inner corners, where small applied loads were concentrated.
The failures caused fires in the vehicles fitted with the defective rubber hoses, and the manufacturer may have (wrongly) relied on an oil film providing protection. In the other case, it was a change of compressor of different design which caused very low levels of ozone to appear in the pneumatic system. The solution to all such problems either means using an ozone-resistant rubber such as EPDM, or adding an anti-ozonant chemical to counteract traces of the gas.
11.5.3 Environmental stress cracking
Unique to polymers, ESC occurs when products are exposed to active organic liquids or even vapours. The problem was encountered on the assembly line where miners’ battery cases were made, cracking occurring when the injection mouldings were exposed to strong solvents being used to make welds between the cases and the lids. The cracks went mainly unobserved by quality inspectors, but appeared of visible size in colliery lamp rooms and at the coal face. Examination of new parts showed high levels of birefringence due to moulding into cold tools, but changing to high temperature tools improved product life greatly. Many other improvements to the design were also made.
The problem re-appeared in the 1990s when polycarbonate connectors used in IV lines in hospitals started cracking when in use and exposing patients to the possibility of bacterial and viral infection. One design failed at numerous separate hospitals, probably caused by contact of the outer surface with common cleaning agents such as bleach, a weakly alkaline liquid which also contains chlorine. However, the inner bore could also be attacked by lipid solutions such as Total Parental Nutrition, a synthetic fluid similar to milk for premature babies. It was inferred that the device had been moulded using cool or cold tools so that there was a driving force for crack growth.
The explosion of a compressed air line provided a dramatic example of the problem. Although no-one was injured, there was substantial physical damage to the glass factory where the accident occurred. The pipes of the system were made from ABS, which was stressed highly by the air inside. The walls were under high hoop stresses, and when organic fluid entered the pipes, it reacted with the inner bore to produce deep crazes and cracks. When one became critical, the pipe blew apart. The fluid probably came from the compressor, in which an unapproved oil had been used.
The problem of ESC is created by surface absorption of the fluid, which swells the polymer to a greater or lesser extent, the degree of swelling depending on the chemical compatibilities of the polymer and the fluid (6). Swollen polymers are weaker than solid bulk, so cracks can be formed quickly and grow when the product is under stress or strained internally by residual stresses or strains. Amorphous materials like PMMA, polycarbonate and ABS are more susceptible than semi-crystalline polymers because the crystallites are much more resistant to swelling. However, they are not immune to ESC, and early use of polyethylene in blow moulded containers, for example, could crack when exposed to strong detergents. We were referred the problem in the 1970s when a number of such containers fractured, releasing large quantities of beer. They were cleaned regularly with a very strong detergent, Igepal, which initiated ESC cracks. They grew with time under the influence of blow moulding orientation until sudden failure occurred. The solution to the problem lay in increasing the molecular weight of the raw polyethylene used in their construction. Higher molecular weight grades are always more resistant to ESC since they provide a greater reserve of strength.
11.5.4 Data compilations
Data compilations of interactions between different polymers and chemicals are frequently provided by material suppliers, usually in the form of an extended table. The information is purely qualitative only, using the terms ‘no effect’, ‘good-minor effect’, ‘moderate effect – fair’ or ‘severe effect – not recommended’, for example. Given the sheer number of the possible combinations, they can be a useful starting point for guidance to potential users or investigators. However, further information should always be obtained by direct experiment under conditions directly related to those of interest. For example, the effects of temperature are either completely ignored or limited to one or two temperatures, so detailed resistance must be determined experimentally.
The tables also ignore the effects on different grades of polymer and, as already seen above, the grade can have a strong influence on cracking. Low molecular weight grades are inevitably weaker than high molecular weight grades. Individual polymers may also show a wide variation in crystallinity depending on their processing history, which is yet another reason for direct analysis. There is usually little information on the effects of changes in concentration of the attacking agent, which is often critical. Dilute reagents will tend to be weaker cracking agents than more highly concentrated reagents for example.
11.6 Access to information
The availability of information concerning the limits of polymers is widely disseminated in the literature and was formerly difficult to access. It was spread between suppliers’ technical data sheets or brochures, the technical literature and textbooks, although much key information remained in company archives, where it never saw the light of public scrutiny. The acetal and polybutylene failure reports made by the material suppliers in the case of the US plumbing class action is an example of such secrecy, and only exposed by their discovery during court action. Of perhaps even greater concern was the way in which those companies ignored their own detailed analyses of the problem.
11.6.1 Published literature
There are problems with all public published information sources, such as the technical literature and textbooks. Because they are paper sources, they have often been edited or even worse, sanitized to exclude significant details. They are frequently out of date, or overtaken by other events, a problem in an area of rapidly developing technology such as polymers. By contrast, metal technology is much more mature, and considerably more information is available to users. Nevertheless, it tends not to discuss problems in terms of case studies, where the context is often critical.
The scientific and technical literature is also mainly directed towards theoretical rather than practical ends, so is limited in the help it can give to investigators. There are some notable exceptions, such as the French paper giving details of the UV sensitivity of thermoplastic nylon elastomers mentioned in Chapter 4. There a few journals that are devoted to failure analysis, most notably Engineering Failure Analysis, where case studies are extolled. Papers dealing with failed products do appear from time to time in other materials or engineering journals, but are widely scattered. Another exception is medicine, where failures are reported in detail, although the detail often excludes engineering information of critical importance. There are detailed case studies published by RAPRA, which has taken a lead in publishing their own files in an effort to make users more aware of the problems of polymer degradation (5). But generally the literature neglects discussion or even mention of failures, probably because there is a systemic aversion to publicizing failed products or processes.
However, such discussion is vital when major accidents or disasters occur, and we are fortunate that the results of public enquiries are freely available. The NTSB and HSE are publicly committed to disseminating such results because such incidents must be prevented by an understanding of the causes. But many less serious incidents are suppressed or published in places less accessible to those who need to know about a potential problem. Most major accidents are preceded by warning signs or less serious incidents, or in some of the worst examples, by previous accidents. Much of the information is revealed after a major accident when a court case arises, such as occurred in the case of the motoway accident in France. There had been several similar accidents, some fatal, when the same design of bag had fallen from the rear of motorbikes and led to seizure of the back wheel. The accidents were in the public domain, but key details seemed to be missing in newspaper reports. When they were all put together with relevant details, however, a picture emerged of a design with serious flaws.
11.6.2 Old problems
One other aspect of the problem of data dissemination relates to old problems solved at the time, but which re-appear many years later when the original problem has been long forgotten. A good example is the problem of ozone cracking, a failure mode long known about and even familiar to car drivers before the use of anti-ozonants in rubber products became widespread. Since vehicle tyres are the most important elastomeric product, use of these protective additives suppressed the problem. But it didn’t go away for many other safety-critical products like fuel lines, where fires continue to occur even after warnings of the problem of radial brittle cracks penetrating the bores, as the recent example of aircraft towing vehicles showed. Although much is known about the mechanism of attack, users may be quite unaware of the problem if they have never encountered it before.
Small but critical components such as diaphragm valves were also affected when the compressors were replaced by a quite different design. The new machines produced tiny amounts of ozone but which were enough to bring an entire semi-conductor line to a halt. The aggressive gas is also more likely to be present under certain conditions in the atmosphere as a result of local conditions, especially when organic pollutants such as petrol or diesel fumes are acted on by sunlight. So the problem is likely to continue when unprotected elastomeric products are exposed to the gas.
The acetal cracking problem is another example of an old problem which reappeared with devastating results. It had been known when two versions of the polymer were commercialized that extra stabilizing methods were needed. One version was end capped to prevent the chain unzipping from the chain ends, while unzipping of the copolymer version was provided by blocking groups of the more stable copolymer unit (7). But the polymer was known at that time to be unstable to hydrolysis, especially in acid solutions, and the fact that chorine dissolves in water to form dilute hypochlo-rous acid will have alerted later workers to the problem. We know that was the case because of the research reports from the 60s and 70s about the degradation of acetal mouldings when exposed to very low levels of chlorine in a water supply. The problem lay in managements who either did not know of their research or who blithely ignored those internal reports. The result was one of the largest compensation claims ever recorded in the USA.
So only publicizing the case to future workers will or should prevent further lapses in judgment. One new factor that suggests that those results will not be forgotten lies in a new way of accessing information: the internet.
11.6.3 The internet
It is the world wide web that has transformed the information problem for the better in the last decade. The growth has been staggering, both in terms of numbers of people who can access the web and the sheer amount of information presented. The main search engine, Google, provides searches of many different types of information, including searches of books and the technical literature as well as images. The latter is especially useful for finding photographs of failed or fractured products, and was used to find several images of exploded batteries in Chapter 5, for example. They showed very similar effects to the examples discussed there. Many personal website blogs give information on failures experienced by their authors, an incidental but valuable contribution for investigators. Such ‘metadata’ (8) can provide a broader picture of failure than is often possible from their own information. Metadata is the information usually deleted by editors from official reports, but is frequently useful in providing key data not accessible in other ways.
Like their corresponding paper literature, company websites praise the positive virtues of their products with little or no mention of the drawbacks or problems. But technical brochures are a good starting point and certainly much technical information is potentially very useful to designers. Case studies in commercial literature always focus on successful products and their development, without giving counter-examples. One notable exception was the ozone attack paper published by a large pneumatic manufacturer in Japan, which unfortunately was not read by one of their competitors! Forensic company websites are exceptional in giving case studies of failed products, but the cases are normally very brief and lack supportive information. They are also mainly restricted to metal products, with non-metals rarely mentioned.
Much legal information such as court judgments and class action announcements are available on the web, especially in the USA. They were the source of some of the discussion of the acetal/polybutylene problems of Chapter 6. Government websites are valuable for the many official reports of major accidents and disasters, and include the NTSB, MDHA (formerly the MDA), FDA and HSE. The FDA website is especially interesting for learning about medical device failures, as was discussed in Chapter 3. All of the recent US government reports of disasters from the NTSB are available for free download, although older reports are difficult to access, presumably because they have not yet been scanned and downloaded to their website.
University web sites are a growing source of information on product failure, especially where staff are active in forensic research. Some sites discuss case studies in detail (9), although many more discuss current theoretical work and other current research.
11.6.4 Wikipedia
One central source of information is the web-based encyclopaedia, Wikipedia. Founded only in 2001 by Jimmy Wales, the site offers an unprecedented source of information on every possible subject (10). As of 2007, there were 2.7 million articles on the English language version, and the number is still growing at a fast rate. It is an open access source of information, which currently means that any user can edit articles, but the policy means that vandalism is a severe problem. There is also a problem of spam, especially in technical articles, where commercial users attempt to insert links to their own websites.
The breadth of the site is awesome, ranging from trivia and ephemera (video game characters, for example) to everything that one might expect in a conventional encyclopaedia like Britannica. Despite the vandalism, the number of mistakes is about the same, but unlike Britannica, those mistakes can be corrected very easily. Ironically, some articles are based on the 1911 edition of Britannica, which is now out of copyright. The mechanics of editing are very easy to grasp, although there are rules to ensure that it is conducted fairly and without bias. Although the quality of some technical articles is very high, the quality of polymer, material and engineering articles is very variable, with many individual polymers barely discussed in any detail at all. One exception is the article for polycarbonate, perhaps because the material is now so widely used across a spectrum of industries. But because Wikipedia is open for anyone to edit, then there is hope of future improvements. Its open access also means that articles are kept up-to-date by its editors, a severe problem of paper-based encyclopaedias.
It is a source of information on product failures and forensic disciplines (fire investigation, forensic identification, trace evidence and so on), some of which we have helped to write and edit, so opening up the subject to wider view. One fascinating feature of the articles is the way in which individual terms or phrases can be linked, so that a browser can follow a subject through many different pathways. It means that users have much wider and deeper access to fundamentals, applications and implications than is possible with a paper encyclopaedia. Each article is also supplied with external links so you can access other independent sources on a specific topic.
There is, however, great scope for improvement in the area of forensics, especially as the amount of information on product failures grows with time and with greater transparency in publicizing poor product performance. There is no doubt that the journal Engineering Failure Analysis has made great strides in publishing revealing and incisive case studies from investigators, and its expansion in recent years reflects the growing desire to publish detailed case studies so as to limit further failures of the same kind, if not eliminate them. It, too, is readily accessible on the web, so that even the most arcane failure problems can be located by search engines using appropriate keywords. The growth of the web encourages minority interests, or ‘long tail’ of information and sources (11), both in commercial sites like Amazon and independent sources like Wikipedia.
11.7 References
(1) Peterson, R.E. Stress Concentration Factors. In Pilkey W., ed.: Wiley-Interscience, 2nd edn, Peterson’s Stress concentration factors, Wiley-Interscience, 1974. [(1997).].
(2) Young, W.C., Roark’s Formulas for Stress and Strain, 6th edn, McGraw-Hill, 1989.
(3) Mohammadi, S., Extended Finite Element Method: For Fracture Analysis of Structures, Wiley-Blackwell, 2007.
(4) National Transportation Safety Board. A Highway Accident Report, Collapse of US 35 Highway Bridge. Washington: GPO; 1971.
(5) Wright, D., Failure of Plastics and Rubber Products: Causes Effects and Case Studies involving Degradation, RAPRA, 2001.
(6) Brydson, J., Plastics Materials, Butterworth, 7th edn, 1999.
(7) Barker, S.J., Price, M.B. Polyacetals, Iliffe Books/Plastics Institute, London (1970); Vogl, O. Edward Arnold and Dekker: Polyaldehydes; 1967.
(8) Weinberger, D. Everything is Miscellaneous: The Power of Digital Disorder. New York: Holt; 2007.
(9) Websites with extensive collections of failure case studies include: http:// technology.open.ac.uk/materials/mem/ and http://www.tech.plymouth.ac.uk/ sme/UoA30/Consulting.htm#Consulting%20Work.
(10) See the entry at http://en.wikipedia.org/wiki/WIKIPEDIA. Editors can track the status of articles on their own watchlist.
(11) Anderson, C., The Long Tail: How Endless Choice is creating Unlimited Demand, Random House, 2006.