Forensic engineering has grown substantially in recent years as consumers have demanded ever-increasing levels of quality. Premature product failure not only deprives the users of that product, but can also lead to personal injury and other detrimental effects. This unit introduces you to a subject of interest to engineers, designers, patent agents and solicitors.
This OpenLearn course provides a sample of level 3 study in Engineering
After studying this course, you should be able to:
analyse the cause of failures in polymer products
recognise safety-critical polymer components in products
describe different failure modes, using conceptual and engineering diagrams, such as fault-tree analysis diagrams
distinguish between non-specification features and defects in polymer products. Suggest ways of reducing the effects of defects
extract key information from standards, specifications, codes of practice, or technical literature and apply that information to product failures.
The subject of this unit is forensic engineering, a branch of engineering that has grown substantially in recent years as consumers have demanded ever-increasing levels of quality. Premature product failure not only deprives the user of that product, but can also lead to personal injury and other detrimental effects. If a ladder suddenly collapses due to the fracture of a key part, the user may be thrown off and injured. If the radiator of a new car suddenly runs dry and the engine seizes up, the owner will want redress. When a patent for a new product is copied by others, the patentee will lose control of his or her invention unless action is taken against the infringer. They are all examples of the kinds of problem to which the forensic engineer can make a neutral and objective contribution.
The subject has grown with the application of new investigative tools such as Fourier transform infrared spectroscopy (FTIR) and scanning electron microscopy (SEM), techniques also used widely by forensic scientists (Box 1).
Consideration of documentary evidence, which accompanies any dispute, is also a task the forensic engineer can perform. So any engineering drawings of a product that becomes available during a dispute may need to be interpreted for dimensional tolerances, or other features that may have caused a problem.
The work of forensic scientists is well known because of high-profile murder trials, where forensic evidence was crucial in determining the truth. DNA analysis has become so sensitive a method that old, unsolved cases have been re-opened with new evidence obtained from traces of tissue still preserved from the original crime scene. While such cases have undoubtedly been spectacular successes, we should not forget that other cases have been equally spectacular failures.
There has been a series of major blunders where innocent individuals have been wrongly convicted on the basis of poor forensic analysis. Many of those cases were associated with IRA bombing campaigns, where misinterpreted forensic results combined with false confessions to produce unsafe convictions. Simplistic tests for traces of oxidized nitrogen – the nitro group – were used on clothing and limbs to show the individuals had handled explosives, for instance in the McGuire and the Birmingham Six cases. In fact, such traces could be reproduced from a wide range of common products, such as playing cards coated with cellulose nitrate, and even staples held together as a strip with cellulose nitrate polymer.
Forensic scientists practice in a variety of fields, ranging from fingerprint detection and analysis, to document examination and chemical analysis of drugs and poisons. The results of their work are widely used by the police in proving physical connection between suspect individuals and the crime scene. They are based mainly at several large and well-equipped sites supported by the government or police authorities, although they accept commissions from defendants who wish to test prosecution evidence independently.
Many of the techniques used by forensic scientists can be valuable to forensic engineers, who tend to work mainly on civil cases involving product failures, and accidents. In disputes involving intellectual property such as patents, forensic scientists are not involved at all, because such cases involve interpretation of the function of machines or devices.
Forensic engineers can help in criminal cases where sample examination and interpretation of a special kind is needed. The scanning electron microscope (SEM), frequently used for the inspection and analysis of fracture surfaces, can prove equally valuable for analysing traces of contaminants on products from a crime scene. Analytical methods such as spectroscopy have reached levels of sensitivity unavailable before 1990. Nevertheless, such methods also need the checks and independent corroboration that are an essential part of thorough sample examination. Such checks include calibration of the analytical machines, and the use of standard specimens to check machine function, as well as protection against contamination of the original samples.
Both forensic scientists and engineers can appear in criminal or civil cases, and as expert witnesses, can provide opinions to courts or lawyers. This privilege is normally denied to witnesses of fact, who can only give evidence of their direct involvement in an incident or crime. An eyewitness can give testimony about an accident she has seen personally, but cannot normally give an opinion as to why that accident happened.
What is forensic engineering? To answer the question, we need to look at the meaning of the words. Consulting Chambers English Dictionary gives the following definitions:
belonging to courts of law;
appropriate or adapted to argument;
of or relating to sciences or scientists connected with legal investigations.
engineering noun the art or profession of an engineer.
engineer noun someone who designs or makes, or puts to practical use, engines or machinery of any type.
So forensic engineering could mean of or relating to engineers connected with legal investigations. But what legal investigations is a forensic engineer concerned with? They could include:
product failure – such as the breakage of a critical part of a product, as in the ladder case just mentioned;
process failure – such as a manufacturing process failing to achieve the intended effect;
design failure – such as the premature failure of all products in the marketplace;
business failure – such as the infringement of intellectual property rights.
Not all such failures lead to litigation. Product failure occurs frequently with consumer products, for example, but resorting to the law is rarely used to alleviate the problem. A new car may develop an unexpected defect, such as a leaking radiator. If covered by a warranty, the seller will repair the damage at no cost to the owner. One common condition of a warranty is that the car is maintained according to the maker's instructions; if not, the warranty may be void.
Failure is reasonably common with a range of products, and there are well-established routes to replacing failed products without resorting to litigation. Components that have a limited life are therefore routinely replaced as a precautionary measure, and others are changed when wear becomes visible, such as car tyres. However, some failures are more critical than others because they may involve safety-critical parts, for example. A safety-critical part is essential to the correct functioning of a product, because if it fails, people could be put at risk.
It is the primary thesis of this unit that all unexpected failures, however trivial they might appear, should be investigated to determine the cause. The more serious the failure, such as with safety-critical components, the greater the need to investigate. If a faulty car radiator causes an engine to seize, the repair cost escalates; but if the seizure of the engine led to a car accident, the safety issue becomes paramount.
It is the techniques of forensic engineering which, when applied objectively to failures, can isolate the cause or causes. The engineer analyses and isolates, then reaches conclusions that lead to improvements in processing, design, or material. Perhaps the manufacturer needs better quality methods; perhaps there was a design defect; perhaps the problem was caused by the user (by not topping-up a car radiator, for instance). With the problem isolated, action can be taken to eliminate it, or warnings can be issued to other customers.
The case studies in this unit deal with polymer products, and show how forensic methods can be used to determine causation. As a relatively new class of materials, polymers have found widespread application in many consumer and industrial products, often in a safety-critical role. New polymeric or composite materials, such as polycarbonate or polysulphone, offer many advantages in a variety of ways, be it in greater design freedom or lighter weight and so on. But there are some disadvantages, especially as experience of long-term use is often absent. Many new polymers were invented within the last 50 years, so their long-term behaviour is difficult to assess.
Knowledge of the failure modes of polymers and composites is a vital part of their safe application, and case studies are an important way of gaining more precise knowledge of their properties. Textbooks tend to ignore many of the realities of product failure, which is unfortunate given the importance of the subject to designers. Textbooks also tend to emphasise theoretical aspects of particular materials, without mention of the practical problems of using those materials for safety-critical components.
Metals and alloys, on the other hand, have a long lineage, so more is known about their safe application, both in the short-term and long-term. Even so, it doesn't prevent failures in products made using those materials. On the other hand, many failure modes are well established, can be recognised with confidence, and appropriate action taken by the design team.
Analyse and discuss the possible cause or causes of the following failures. Indicate any possible causal connections. Indicate which involve a safety-critical component.
A car fails to start on a cold, wet morning.
The front door on a washing machine opens during a wash, spilling the contents into the kitchen.
An electric kettle boils dry and melts the casing.
A petrol pipe near the fuel pump of a car with the engine on leaks petrol, and the car is destroyed by the ensuing fire.
(a) A car that fails to start on a cold, wet morning could be caused by several factors: if petrol fuelled (rather than diesel), the electrical system may be the cause. It is well known that moisture deposited on cold metal or ceramic surfaces can allow leakage of charge from the high-voltage circuit supplying the plugs. Alternatively, the petrol tank may be empty, or the car may need servicing. The battery charge may be too low to turn the engine. There are clearly a number of possible causes.
(b) The failure of the front door of an automatic washing machine during a washing cycle suggests the door catch may have failed – if the door is still attached by its hinges. Such latches are usually strong, because they are safety-critical, and are locked at the start of the washing cycle. If the complete door fell away, both hinges and latch have failed to fulfil their design function. Hinges are also strong and safety-critical, and there are many possible reasons for failure, including fracture – the complete or partial separation of a key component. Alternatively, the door may not have been firmly closed by the user of the machine. Another possibility is that the door appeared latched and locked, but was not, due to some fault with the control mechanism of the door.
(c) An electric kettle is a familiar enough device in the home, and it relies for its safe functioning on a thermostatic switch that will turn the current off when the water boils. The switch clicks audibly when this occurs during boiling. If the water boils dry, however, it is reasonably clear the switch failed in its primary function of controlling the electricity supply.
(d) A fuel pipe conveys petrol from the fuel pump to the carburettor or injectors, when the engine is turning. If it fails and fractures petrol will leak, and owing to its proximity to a hot engine with numerous electrical components, can easily cause a flame or even an explosion. Fracture can occur by many mechanisms, but one obvious failure mode involves overloading. It is clearly important to determine the material of construction because different materials behave in quite different ways when stressed. Only a tough material that is crack-resistant should be used, but a mistake could allow a brittle material to be used, or the originally tough material may have become brittle in the engine compartment. The engine compartment is hot and there are numerous chemicals that can escape, such as battery acid, water, antifreeze, petrol itself and various oils and hydraulic fluids. They can react with fuel pipe materials if circumstances permit.
All the above examples of product failure can be safety-critical, depending on the circumstances surrounding the failure. If the car was parked in a heavily used road, failure to start could be safety-critical. Likewise, failure of the door of a washing machine could be safety-critical if the ensuing flood reached electrical circuitry causing a short circuit and hence a fire. Failure of the kettle control switch could also be safety-critical if the casing melted and caught fire, and the fire spread. Live wires could be exposed during the failure, creating another hazard. The car fire is clearly a danger to the occupants.
The examples given in the previous SAQ illustrate a central feature of failure analysis: an implicit assumption that there are rational causes for failure, and that there are reasonably objective ways of determining those causes.
It is assumed you have some inkling of the way common machines operate. This is not going to be true of all machines, even in the home. Microwave ovens, personal computers and CD players come to mind.
With a fault on a domestic machine I expect you would look for the operating manual, which gives basic guidance on common failure modes, and how to recognise their symptoms, often in the form of a fault-finding guide. Typical fault-finding guides from car maintenance manuals are tabulated and suggest possible causes for one specific symptom.
For the cooling system part of the fault-finding guide shown in Table 1, five general symptoms are listed, and possible causes listed under each heading. It is then for the mechanic to investigate each possible fault mode systematically in turn to find the faulty component. By eliminating parts that are functioning correctly, the faulty part can be located. As internal-combustion engines work on the same general principles, the tabulated data is generic to different makes of car. Such checklists are certainly a useful start to troubleshooting, but must be used with common sense, so that unsuspected failure modes are not discounted.
One simple way of showing the inter-relationship of many different possible causes producing a single symptom is known as fault-tree analysis (FTA) (see Figures 1 and 2). An overheating car engine can be caused by insufficient coolant (Table 1, line 1). This, in turn, could be caused by internal or external leaking of the coolant. An internal leak could be caused by a broken cylinder-head gasket or a cracked cylinder head, for example. Such internal leaks usually lead to water entering the oil sump, which turns the oil milky, and is instantly recognisable. Alternatively, the coolant could leak to the exterior because of a broken hose, damaged seals or a faulty pressure cap. You will notice there are several dotted lines on the fault tree, which indicate other possible but unknown causes that might be peculiar to a model of car.
|Cooling system faults|
|overheating||external coolant leakage|
|□ insufficient coolant in system or thermostat faulty||□ deteriorated or damaged hoses or hose clips|
|□ radiator core blocked or grille restricted||□ radiator core or heater matrix leaking|
|□ electric cooling fan or thermoswitch faulty||□ pressure cap faulty|
|□ valve clearances incorrect||□ water pump seal leaking|
|□ pressure cap faulty||□ boiling due to overheating|
|□ ignition timing incorrect/ignition system fault – petrol models||□ core plug leaking|
|□ inaccurate temperature gauge sender unit||internal coolant leakage|
|□ airlock in cooling system||□ leaking cylinder-head gasket, cracked cylinder head or cylinder bore|
|□ thermostat faulty||□ infrequent draining and flushing|
|□ inaccurate temperature gauge sender unit||□ incorrect coolant mixture or inappropriate coolant type|
Fault-tree analysis diagrams are useful because they place analysis on a systematic footing, and give a mechanic a logical path to follow during examination. Each failure will exhibit well-defined features. If inspection of the engine near to a cooling hose shows one point where there are dried traces of impurities from the cooling system, the adjacent hose will need closer examination.
With machines where the working principle is unknown, it is often possible to evaluate failure using common sense. As you might expect, the investigation becomes a logical analysis of the way failure occurred.
If the original failure is buried under consequential damage, the task is more difficult than for simple failures. The failed component may lie deep within the machine, and the failure only inferred from the lack of a particular function. In addition, further clues to the failure may be exhibited by contact traces of wear or abrasion for example, where none would be expected. Analysis of such traces is known as trace analysis or contact analysis, and can be extremely useful in locating the source of a particular problem, as you will see later in the case studies. Forensic investigation often starts after a mechanic finds a component that should not have failed or shows defective design or manufacture.
Table 2 shows the checklist of faults for the fuel and exhaust system of a car. Construct a fault tree to show causal relations between excessive fuel consumption and various car components. Feel free to add any other possible causes to your diagram not shown on the checklist. Include possible safety-critical defects. How would excessive fuel consumption be detected by the driver – rather than by the mechanic?
|Fuel and exhaust systems faults|
|excessive fuel consumption||fuel leakage and/or fuel odour|
|□ air filter element dirty or clogged||□ damaged or corroded fuel tank, pipes or connections|
|□ fuel injection system fault – petrol models||excessive noise or fumes from exhaust system|
|□ faulty injector(s) – diesel models||□ leaking exhaust system or manifold joints|
|□ ignition timing incorrect/ignition systems fault – petrol models||□ leaking, corroded or damaged silencers or pipe|
|□ valve clearance incorrect||□ broken mountings causing body or suspension contact|
|□ tyres under-inflated|
Figure 3 is my attempt to show causal relations between excess fuel consumption and faulty components. However, excessive fuel consumption could also be caused by driving conditions, because excessive starting and stopping uses up more fuel than motorway cruising, for example. The checklist shown with the question also indicates fuel leakage as another cause. As this is potentially a safety-critical defect, it should be included in the fault-tree diagram.
Detection would occur by careful monitoring of fuel consumption and mileage driven. However, some new cars have automatic monitoring systems, so that the driver is warned of any consumption problems. The driver would also normally check tyre inflation pressures, but the other possible causes shown in the checklist would normally be performed by the mechanic during servicing.
Table 3 shows the fault-finding checklist for the manual transmission system of a car. Construct a fault-tree analysis diagram for low oil levels in the sump as detected by an oil warning system, gauge, or dipstick. Add any further possible causes from your own knowledge of internal combustion engines. Briefly indicate the role of engine oil in a working engine. What are the consequences of low oil levels to engine performance?
|Manual transmission faults|
|noisy in neutral with engine running||jumps out of gear|
|□ primary shaft bearings worn (noise apparent with clutch pedal released, but not when depressed)||□ worn or damaged gearchange linkage/cable|
|□ clutch release bearing worn (noise with clutch pedal depressed, possibly less when released)||□ incorrectly-adjusted gearchange linkage/cable|
|noisy in one particular gear||□ worn synchronizer units or worn selector forks|
|□ worn, damaged or chipped gear teeth||vibration|
|difficulty engaging gears||□ lack of oil|
|□ clutch fault||□ worn bearings|
|□ worn or damaged gearchange linkage/cable||lubricant leaks|
|□ incorrectly-adjusted gearchange linkage/cable||□ leaking differential output oil seal|
|□ worn synchronizer units||□ leaking housing joint|
|□ leaking primary shaft oil seal|
Figure 4 is based on Table 3, especially the final set of checks for lubricant leaks. Another possible cause that has been added includes lack of adequate maintenance (oil replacement at regular intervals, topping up when needed). In addition, the engine oil itself may be of the wrong grade or contain volatile compounds that volatalize easily, or burn in the engine easily because of internal leaks.
The engine oil is an essential lubricant to the various moving parts of the engine. It is normally pumped onto and around such components, so any loss of oil level can deprive moving parts of protection. If too low, metal will move on metal, so creating high friction and wear, with the possibility of seizure.
Manufacturing is about making products to a specification, which is often a standard, such as one published by the British Standards Institution. It is useful to consider the steps that result in a finished product (Figure 5).
The core phases are shown in Figure 5. The early stage of design will establish what environment a product will be used in, and the length of time a customer will use it – its expected lifespan. Product conception is followed by planning and modelling, leading to a product specification. Provided the model achieves the desired technical performance, detailed consideration of materials follows. Testing and evaluation lead to a prototype. Through all the design stages, the failure of models and prototypes is an essential part of the development process, leading to improved prototypes. The process is an iterative one, where failure at any stage causes reconsideration of the specification or materials. Such failures are of course a normal part of new product development.
When a successful prototype emerges, production can be planned. Production machinery is costly, so this stage demands careful planning and testing.
It is after the product is sold into the open market, that product failure looms large in the design equation (Figure 6). Product quality will be a touchstone for public perception of the product itself, so any failures need immediate attention by the design team. Failure investigation using forensic methods is the key to identifying any design flaws or production defects, and correcting them quickly.
One tool for finding and eliminating product defects of any kind – due to design, or materials, or manufacture – is known as Pareto analysis (Box 2). It groups product defects in terms of their importance, so the investigators can gain an overview. The team can then concentrate on the most serious defects, using fault-tree analysis (Box 3).
Another technique of great value in tackling product defects is known as failure modes and effects analysis (FMEA). It is in widespread use in many industries, being required by many standards and codes of practice. It is based on an assessment of the real or suspected defect using three criteria: occurrence, severity, and detection. As it is a key tool for designers working both on new products and products of long standing, it will be described in more detail below.
Failure analysis assumes the design team will have already identified the ways a product can fail. If not, it will be first task of the team to specify the way it could fail, probably by starting with failures during testing of prototypes or models. A tool that can help concentrate minds is Pareto analysis as shown in Figures 7 and 8. It is said it can give an 80 per cent improvement from dealing with 20 per cent of the defects: the 20 per cent made up from columns 1 and 2.
Pareto analysis of defects helps to pinpoint the areas to concentrate on first. Figure 8 identifies the ‘vital few’ being the large numbers of components with defects 1 and 2, and the ‘trivial many’ as the smaller numbers having defects 3–7. It suggests that by reducing the incidence of defects 1 and 2 by 50 per cent, concentration on the vital few will yield the greatest returns for expended effort. But minor defects cannot be ignored if they are safety-critical. The philosophy taken at face value fails to grade defects into their criticality. If, for example, lack of concentricity means a machined part does not fit another, it is far from trivial because material is wasted. More seriously, if safety-critical defects are ignored – as ignoring defects 3–7 might indicate – someone could be injured or killed.
The problem is a dilemma for company managers, because it is short term and focuses on production costs, but in the long term it could lead to consumer dissatisfaction, even injury or death. A number of US car companies have been criticised by US courts in product liability suits, where company accountants have resisted recalls or remedial work on the grounds that the cost would exceed the costs of court awards to injured customers. The first and most notorious case is the Ford Pinto car, which involved petrol tanks that blew up in rear shunts.
Fault-tree analysis when applied in a manufacturing environment is often called cause-and-effect analysis, and results in a cause-and-effect diagram with added cards (CEDAC). It was developed by Sumitomo Corporation in Japan as part of a quality improvement programme, and involves constructing a fault tree. The diagram is like the fault-tree you saw in Figure 2 turned from 12 o'clock to 3 o'clock: the ‘effect’ (the fault) is on the right, contributing causes lie to the left in a fishbone pattern (Figure 9).
Such a diagram is exhibited in the factory and workers can add new possible causes to the diagram – the added cards. The problem is therefore under continual review.
A CEDAC team performs the review, by selecting the most likely causes and testing hypotheses against test data. The goal is to find the root cause or causes of a particular problem, as shown in Figure 10.
Corrective action can then be taken to alleviate or eliminate the problem. The method is in widespread use in manufacturing industry, often in combination with FMEA and other total-quality methods.
A characteristic of many products is illustrated in Figure 11, where failure rate is plotted against product lifetime. The so-called bathtub curve shows an early, high mortality (known sometimes as infant mortality), followed by a stable mid-life of random failures, followed by a rise in failures as critical components wear out.
For safety-critical products, any early failures are unacceptable, so all products will be inspected and tested for their key functions before being sold.
Wear-out failures can be avoided by replacing components likely to degrade at pre-determined intervals. Drawing on the car as an example, some of the components replaced at intervals would be:
cooling system seals;
elastomeric hose for coolant, fuel, and brake fluid;
A sensible maintenance policy is to replace sensitive components at regular intervals. For instance, it is a legal requirement in the UK that tyres be replaced when the tread depth reaches 1.6 mm.
For safety-critical components, a regime of testing every component in the factory can nearly eliminate early failures, and reduce random failures. Wear-out can be reduced by replacing parts at a predetermined point in their lifespan. These measures produce the ideal curve shown in Figure 12 (using the same axes as in Figure 11). However, if parts replacement is botched, safety can be severely compromised, as Box 4 relates.
Seals in closed fluid systems are often safety-critical components because they prevent escape of the fluid into the outer environment.
An astounding incident was reported in the House of Commons in November 1997, and related to the near disaster that occurred to a BAE 146 aircraft of the Queen's Flight. The four engines cut out one-by-one, with the final engine cutting out just as the pilot landed the aircraft.
Investigation of the engines after the incident showed the engines had experienced a loss of oil pressure, leading to shut-down. None of the sump drain plugs on the engines had been fitted with their O-ring seals, so oil leaked out slowly. It was fortunate the last engine was working during the landing, and the aircraft was not carrying any member of the royal family at the time. No doubt the mechanic was duly admonished. The story is a useful reminder of the safety-critical nature of many seals.
Another example of a seal problem arose in the early 1990s when the hot radiators in numerous hospitals and old people's homes leaked and caused material damage. The radiators had only recently been fitted, and incorporated a new material for the flat circular washers used on the drain plugs. The material was a thermoplastic elastomer: a copolyester with polyether rubbery chains. Such polymers offer advantages because they can be injection moulded quickly and easily, when compared with conventional cross-linked rubbers. They also apparently offered better water resistance than the older fibre washers, and had been used successfully in hot water taps.
The initial report to loss adjusters pointed to two possible causes:
overtightening by the plumber;
undertightening of the drain plugs.
The possible causes of failure on such a large scale had not been considered. While either cause could explain a few failures, it did not explain the many installations by many different plumbers. In addition, the initial investigator ignored key evidence of brittle cracking and hardening of the material.
The conditions of exposure are quite different in radiator seals, compared with hot water taps. Washers in the latter are exposed only intermittently to high water temperatures when a tap is opened, while radiator seals are exposed almost continuously. Examination of the failed washers (Figure 13) showed radial brittle cracks through which the pressurised hot water could leak. The material showed little or no resilience, indicating that crystallisation had occurred, making the polymer hard and unyielding.
The investigations carried out independently by several experts showed the material supplier had printed a technical note to warn customers of the instability of the polymer at high water temperatures. For whatever reason, the note had not been read by the moulder, and it had made millions of the washers for the radiator company.
The civil case was eventually settled for a large sum to compensate for the material damage caused by the leaking water, and for replacing the washer material with another polymer, EPDM rubber.
Surprisingly, this example is not that unusual, because new technology or new materials often create problems of use and application that were not foreseen by their originators. However, most manufacturing companies try to explore the properties and potential problems of new materials before launching a product, both by laboratory testing and by development phases such as prototyping.
What would you say are common defects found in products? You might answer the question by asking what is meant by a defect. A working definition might be:
a defect is a feature of a product or component that inhibits or prevents the correct operation of that product or component.
The term feature is here used in a neutral way to denote any characteristic that does not normally appear in a specification, such as included in an engineering drawing. Variations in surface texture often appear on plastic mouldings or cast metals as a result of the manufacturing process, but designed into the product.
The definition of features that are also defects is useful because it allows us to examine those features that can affect product function in a little more detail.
The very form of the bathtub curve suggests there are at least two types of defect: primary defects, and defects that develop with time. Primary defects are already present before sale and lead to rejection or functional failure. The second type causes failures after considerable use. (See Table 4.)
|Primary defects||Secondary defects|
|Design: geometry||dimensions (overfit, underfit) stress raisers||cracks from stress raisers|
|Design: materials||poor materials||degradation|
|Manufacture||poor moulding, casting etc.||distortion over time|
|Assembly||poor fit||wear of parts in contact|
|poor welding etc.|
|Finishing||surface blemishes||degradation on surface|
Primary features that are often defects in components would include:
sharp corners in stressed areas;
internal voids or inclusions that overstress the product when it is loaded;
deviations from proscribed dimensions so that the component doesn't fit with others during assembly, perhaps caused by the sinking of surfaces of castings or mouldings;
poor quality material, perhaps caused by contamination;
cracks in components.
Any such features can occur in combination with one another, and may also interact: a void near a sharp corner would be a serious defect in a product where the stress is at the corner. Not all such features may be defects by themselves. Whether a feature can be regarded as a defect depends on the product specification.
Features of the second kind, which occur after considerable use, are:
wear of parts that move against one another;
cracks that develop over time or cycles of use;
corrosion by contact with the environment;
change in dimensions with time;
material changes with time.
Like primary features, they will become defects if they lead to fracture or failure, or rejection for aesthetic reasons. The distinction between product features and defects is closely related to the product function, its environment, and usage.
Some severe defects are caused by stress concentration at sharp corners, or from cracks, or from impurities or voids within materials. When products are loaded as part of their normal function, the stress at such defects is raised. These points, called stress raisers, cause the local stress to rise above the intrinsic strength of the material. Cracks are initiated at such points and can either grow rapidly, producing a sudden fracture, or can grow step wise in a manner known as fatigue.
The sketches in Figure 14 show a polysulphone sight tube in outline. It is a part of medical apparatus for controlling airflow to patients. The central part of the sight tube is transparent so a metal float in the centre section can indicate the airflow within (see Box 8).
Which of the features shown are potential defects, and how could they make the product defective? By discussing their possible source and origin, indicate possible ways to eliminate them.
Sight tubes are made by injection moulding, which involves using granules of virgin polymer fed into an externally heated barrel. The moulding screw that fits inside the barrel rotates at a regular rate to homogenise the molten polymer, which is injected under pressure at regular intervals into a heated metal tool, where the polymer solidifies to create the product shape.
Three of the features in the sight tube shown in Figure 14 can be regarded as defects if they make it harder to see the interior flow indicator, so they would affect its function. Another reason for classifying the first three as defects is that, being visible to the user, they may be judged as outside the specification as expressed on an engineering drawing of the product. The final feature, sink marks below the clear central section of the tube, could only be classified as defects if they affected the flow of air through the tube. They are not as visible as the previous features, so might be acceptable to the customer. On the other hand, they would not appear on any engineering drawing, so are strictly outside the specification.
The remedial measures would be as follows:
Foreign particles. The remedy depends on the source of the contamination. The virgin granules could be contaminated, so it would be essential to ensure the granules were completely clean before being fed into the barrel. The feed to the barrel should thus be guarded from the environment. Another source might be wear debris from the machine itself; this could be prevented from entering the product by putting a mesh ahead of the molten stream.
Particles of unmelted granules. They show the granules are not being homogenised completely into the melt. An obvious remedy is to ensure the barrel heaters are working correctly, and ensure the screw is turning fast enough to mix the material.
Scratches. These defects could be created in several ways. They may be due to handling after the solid product has been removed from the tool, so handling would need examining. Gloves might be needed. A more troublesome source could be scratching of the tool itself, in which case the scratches will be identical from component to component. Polishing the metal tool is the remedy.
Sinking of the inner tube. This could be created by under-pressurisation of the melt, so the working conditions of the process might need re-examination if the features are regarded as defects.
Owing to the high cost of testing every product, it is rarely used for mass-manufactured products. Samples are usually chosen to be representative of the batch (as used, for example in statistical process control, referred to below). Tests are normally carried out under standard conditions, defined either by a national or international standard, or by an internal company standard. While they may provide a degree of quality control, the aim of standard tests is to simulate conditions of use of the product. If the test fails to detect product defects, faulty products can enter the market. Failures of the test method itself can occur in several ways.
Failure to reproduce conditions of use.
The test technique is not monitored correctly.
The test may not adequately detect product defects.
There could be the complete absence of testing under realistic conditions, perhaps because wrong assumptions are made, or because of unfamiliarity with a new material or a process. Box 4 relates how failure to test adequately led to a widespread product liability problem. Other examples will be considered in later case studies, both in this and later blocks.
A central heating system operates for long periods at temperatures around 70° C. Figure 15 shows the variation in the half-life of two thermoplastic elastomers with temperature and time of exposure, due to degradation by hydrolysis under test conditions. Calculate the time for the initial strength to drop by half for each elastomer. Which grade is more resistant to the effects of high temperature if used for a radiator seal? What will be the effect of:
turning the heating thermostat down by 10° C?
turning the system off every other day?
Assume the effects of hydrolysis on the material are cumulative.
Figure 15 shows the inhibited 72D-with-additive grade is more resistant to the effects of degradation than the alternative. By reading the graphs, the half lives of the elastomer seals at 70° C are:
72 D with additive, approximately 1.5 years;
72 D, approximately 40 days.
The effects of lowering the water temperature by 10° C to 60° C will be to increase the half lives of the grades to:
72 D with additive, approximately 2.5 years;
72 D, approximately 80 days.
Switching the system off every other day would double the lifetime of each grade, so:
72 D with additive, approximately 5 years;
72 D, approximately 160 days.
An important tool for considering potential and actual failure modes in products is failure modes and effects analysis (FMEA). It is not in itself a product test method but a way of assessing product defects. The results could change product testing or even bring about new test methods.
Pareto analysis, fault-tree analysis (FTA), and statistical process control (SPC) are all techniques used by design teams both for new and existing products. Factories gather data on their product lines, and quality departments use that data. Many inspectors are not engineers, however, and may not recognise a particular problem as rating highly in importance compared with others. For instance, surface blemishes are easy to spot on the line and are clearly defects of a kind, but not usually serious enough to affect product safety. Such patent (obvious) defects are in contrast with latent (hidden) defects, which lie within products or components. The study of patent and latent defects is the responsibility of the design team, which will use FMEA to identify, to classify and to act. They may be defects in products returned from customers, or found within the factory by routine testing and inspection.
A common design defect in many products is sharp corners, which when stressed, raise the local stress above the failure stress of the material concerned. Fracture will start at such corners, and propagate into the interior of the sample. If the product shape is made by a mass-manufacturing route such as injection moulding or casting of a metal alloy, then it will occur in every product so its likelihood of occurrence is 10, on a scale of 1 to 10. If the severity of the consequences of fracture is high – say 8 out of 10 – and it is easy to detect – say 8 out of 10 – the product of all three factors is 640. This number is known as the risk priority number (RPN), so
where L is the likelihood of occurrence, S is the severity, and D is the detectability.
The design team assigned to an FMEA analysis will use experience to set a threshold value above which action must be taken. Suppose in this case, the RPN was set at 250. The team must now discuss ways of eliminating the defect, which here is easily done by modifying the tooling to increase the radius of curvature at the corner in question.
FMEA is not the only answer to improving quality. It is a method that focuses the minds of designers onto product quality in a systematic and rational way. Potential problems can thus be tackled at an early stage in design development, rather than in a panic when failures come back from customers.
A case study of the method is described in Paper 1. You should read the paper before tackling SAQ 6. In addition, the important international standard IEC 812 is presented in Paper 2. It describes the method as it would be applied to electrical equipment, but it can be adapted for other engineering products. Papers 1 and 2 are attached as a PDF documents which should be printed out (if possible) to gain the maximum benefit from SAQ 6.
Click on the 'View document' link below to read Paper 1.
Click on the 'View document' link below to read Paper 2.
Paper 1 above describes the application of FMEA to the problem of stapler and tacker design in the context of a total quality management programme. Summarise briefly the way an FMEA study is carried out for a specific product. What defects were found in the product mentioned in the paper? What factors were important in the way the FMEA committee was structured and operated?
Imagine that in a subsequent committee meeting, a study of a new design of office tacker agreed the following variables for sharp corners near the most highly strained part of the plastic cover.
Likelihood of occurrence: moderate failure rate.
Severity: failure by brittle cracking causes the tacker to behave intermittently.
Detection: fault may be noticed by the customer.
Using the tables provided in Paper 1, what action should be taken by the FMEA team?
Paper 1 describes the first step to be taken by an FMEA committee as being to define the device or process under study. The committee is composed of individuals from different departments within the factory. The committee decides on the three factors of a particular product or process failure mode: the likelihood of occurrence of a particular failure mode; the severity of the failure; and the ease of detection of the particular defect involved in that failure mode. The three numbers are multiplied together to give a risk priority number (RPN). If the RPN falls above a threshold value, action is needed.
The specific example given was a new design of cover cap for an office stapler. It had been copied from an older design, but the old design gave problems on the production line, making hand assembly difficult owing to a tight interference fit between the plastic cover and the steel cover plate. Assembly workers had to wear plasters to protect their hands as a direct result. With a risk priority number of 160, the value was greater than the critical value of 120, so action was taken to eliminate the tight fit.
The committee operates at the design evaluation stage, so that paper changes can be made before production tooling commences. An important step is choosing the committee members. The membership should be four-to-six representatives plus the facilitator, with both engineers and non-engineers as members. The facilitator runs the team, organises meetings, and ensures all the vital data is assembled. The facilitator also manages the computer database; the computer itself provides a focus of attention to prevent cross-talk – talk at meetings that can deflect members from the important subject.
In the case of the sharp corner on a moulded tacker cover, the values for the factors are in the tables at the end of the paper.
Likelihood of occurrence: 5.
Hence, the RPN would be 405, so action must be taken by the team to ameliorate the problem. In this case, increasing the corner radius by modifying the metal tool would be appropriate.
Although the majority of product or process failures are resolved by on-the-spot inspection, there nevertheless remains a core of failures that cannot be easily resolved without litigation. One party either cannot believe, or just does not accept the claims of the other side – or sides, for multi-party disputes. There may be several reasons.
The evidence is unclear or ambiguous.
The evidence was destroyed by the accident, or was not preserved afterwards.
The evidence may offer conflicting interpretations of what happened.
The failures may be widespread and thus more serious than isolated incidents – typical of a fault not dealt with during the design phases.
Process failures may be endemic – typically, contract work by one company for another is disputed.
Commercial disputes sometimes result in litigation, where a court is asked to examine the facts, and interpret the relevant law concerning the dispute. There are several tribunals available for commercial litigants, as discussed in Box 5.
Some commercial disputes involve the interpretation of the law itself, and the final decision becomes part of the case law in that area. A case might, for example, be the first to be brought under a new statute – an act of Parliament – and the judgment could be important for what the wording of the act means when applied to real situations and problems. Such test cases are especially important in intellectual property disputes, where the current law in many areas of technology is being outstripped by events. Computer software development is so fast these days it is difficult for the law to keep pace, for example.
There are numerous courts in England and Wales to which claimants can turn for redress, as shown in Figure 16. The system comprises two quite separate and distinct areas, the criminal and civil courts. In the former, the state prosecutes individuals, and sometimes companies, which have transgressed the criminal law. The civil courts allow aggrieved companies or individuals to pursue claims against others. If someone injured by a faulty product wishes to claim against the manufacturer, they pursue the manufacturer in the civil courts.
The civil courts comprise the small claims courts, the county courts and the High Court, in ascending order of the value of the claim made. The disputes covered by these courts are known as small claims, fast track and multi-track cases.
There are also special courts for dealing with broadly similar types of claims. For instance, intellectual property disputes, which are usually brought by one company against another, are normally dealt with in the Patents Court. It is part of the Chancery Division of the High Court (Figure 17). Other important disputes involving engineering or technical issues may be brought in the Commercial Court, the new Technology Court, or in the Queen's Bench Division of the High Court.
Although claimants can bring their own cases, most cases are actually brought by lawyers: solicitors who organise a case, and barristers who argue in court before a judge. Most cases end with a judgment in the court of first instance, but if the losing side feels the judgment is unfair or based on misinterpreted evidence, an appeal can be made to the Court of Appeal. Further appeal can be made, with the permission of the court, to the House of Lords. This may occur if there is an important principle of law involved. Nowadays, further appeals can also be made to the supranational European Court.
One key difference between the criminal and civil courts is that criminal cases require proof beyond reasonable doubt – a 99.9 per cent certainty – while civil cases require only that a case be proved on the balance of probability – a 51 per cent certainty. In other words, the onus of proof is much higher for criminal prosecutions when compared with civil actions. Another key difference is that criminal cases are normally tried before a jury, while civil actions usually occur before a single judge.
Criminal actions start before a magistrates' court, and most minor actions are settled there. A small proportion involving serious crimes proceed to the Crown Court. Other tribunals include the Coroner's Court for investigating sudden deaths, public inquiries for major incidents and so on.
Disputes are resolved by trials, although a high proportion are settled before trial as the facts become clear after expert investigation and review. So what is so important about a trial? A trial is a public forum where the opposing parties in a dispute confront one another, and attempt to convince a court of the rectitude of their own case. It is often the first time for confrontation, or at least the first time when others are present to order and marshall all the evidence in a systematic and fair way (Box 6).
Much research is done well before the trial, providing evidence to present to the court. The evidence leads to the preparation of arguments to be used for one side or the other. The arguments are prepared in the form of pleadings, documents presented by each side in support of their case, and prepared by the lawyers with the aid of experts if technical issues are at stake. Frequently, so-called skeleton arguments are presented as more evidence comes to light with the trial approaching, and the original pleadings must be revised.
The forensic engineer plays a key role in marshalling the evidence, both the real evidence – such as failed products – and the often voluminous documentary evidence. For instance, if a product fails, there will normally be in-house documents from the manufacturer relating to quality control. Such documents are usually ‘discovered’ well before trial, and in technical cases may include the following.
Maps, plans, and engineering drawings.
Quality control records.
Models and prototypes.
Specifications and standards.
Such documents assume great importance as a trial approaches, because they help to flesh out the background as well as providing key details about the product in question. Where a claimant has suffered loss or injury from a product failure, he or she does not have the information about the way the product was made, or the way its quality was checked after manufacture. The claimant will often have an expert report on the failure, however, which will point to areas that show how the defect could have arisen or how it slipped through quality control procedures. The expert performs a valuable role here, by pointing out what ‘discovery’ should be made to clarify why something failed.
The raw evidence available for a court to consider comes in several different forms. For technical cases, it usually includes one or more of the following.
Real evidence, which could include the failed product or products, tool parts under question, models and prototypes.
Testimony from witnesses of fact relating to an incident or incidents.
Testimony from expert witnesses.
Pleadings in the case, which have been drawn up by the lawyers for both sides.
Documents discovered from both sides related to the facts at issue.
The trial opens with a speech from the claimant's barrister, which outlines the case against the defendant. It is usually short, and is followed immediately by examination of the first witnesses of fact for the claimant, possibly including the claimant herself. This is usually started by the claimant's barrister, so the questioning is friendly, being designed to put the witness at ease, and elicit favourable information about the claimant's case.
Cross-examination by the defendant's barrister is searching in nature. It is where the work of the court really begins, because the testimony is scrutinised for consistency, ambiguity and sometimes deliberate falsehood. The latter is rare in civil actions, but common in criminal trials. Sometimes cases collapse at this point if the claimant contradicts her own previous evidence, perhaps from a witness statement, and cannot explain the discrepancy.
In a recent case where I was to appear as an expert for the defence, this is just what happened when the claimant contradicted himself on the witness stand, even when the questions from the cross-examiner were put fairly and calmly. The latter had intended only a few questions, because the case hinged on expert evidence, but the questions came thick and fast as his evidence began to unwind. The case ended the next day, because the claimant faced yet more searching cross-examination, and his credibility before the court was by this time rather low. In addition, the cross-examiner had been using the claimant's numerous documents to frame his questions. The judge had scrutinised many and found them lacking key details (some pages appeared on the face of it, to have been deliberately obscured during photocopying). The judge asked for the originals, but they were not produced by the time the case collapsed. The defendant dropped the counter-claim, and won all the costs.
The end result could not have been predicted before the trial started, but demonstrates the effectiveness of cross-examination to reach the truth of the matter in dispute. The experts were never called, and the technical merits of the case never tested. It is not uncommon in full trials for them to halt mid-way, and it is even more common for civil trials to stop at the doors of the court when final offers are made between the opposing lawyers. It is a crunch point for both parties, because they know all the evidence will be tested in minute detail. Any weaknesses will be exposed to the light of open court, held in the public eye. So both sides face reality before the trial commences, and often reach a compromise without adding the expense of a trial. Other ways of resolving disputes include arbitration and mediation, collectively known as alternative dispute resolution (ADR).
But it is the material evidence that usually forms the starting point for a forensic investigation, and it is often that evidence which provides a more detailed picture of what actually happened. In an accident, the evidence often gives vital clues about how it was initiated, perhaps from traces of contact with other bodies, or from features that reveal the way the product was made originally. Analysis can also show whether the material of construction was appropriate to the function it performed in service; whether it was affected by processing during manufacture; or whether it was affected by something in the working environment. It is the role of the expert engineer to carry out such investigations (Box 7).
The forensic engineer plays an important role in the resolution of commercial and non-commercial disputes involving technical evidence. For instance, if the failure of a product or component has caused injury, analysis of the failure is critical to the claim for compensation. He or she will be given the failed product, as well as background information on the accident, such as witness statements, and instructed to prepare a report that should include:
the nature of the failure and its consequences;
analysis of the failed product itself;
interpretation of the evidence of the failed product;
possible causes of the failure;
the most likely cause of the accident.
The expert may feel some responsibility to the solicitor paying for the service, or if an expert within a company, some loyalty to that company. However, even if the case never reaches court the report should always be completely objective, and all attempts to bias the conclusions one way or another from outside pressure, whether real or imaginary, must be resisted. A report must be, and be seen to be, a fair attempt to explain a failure in a neutral way. There are several ways of achieving this objective, but the most important is to let the real evidence speak for itself. In a way, a piece of real evidence is a silent witness that will reveal all kinds of detail about the failure that may not be immediately obvious to the untutored eye.
What does a typical report contain? Photographs of pertinent details are normally presented in the report, together with an analysis of the material of construction, an account of any mechanical tests on the material, evidence from new, intact products, and so on. A picture of the sequence of events will emerge as the evidence from the failed product builds up in the report. Finally, a reconstruction of events allows the investigator to re-visit the facts that have been established, and so draw conclusions about the cause. The report should say how the product was made, and whether the material and processing were well chosen.
Much information will necessarily be missing at this stage of an investigation, so the report can point out what extra information is needed from the manufacturer to test the conclusions reached in the report. If for example, the product failed because of faulty manufacture, the report will list the internal documents needed for the process of discovery, a legal process initiated by the instructing solicitor.
On the other hand, if the product failed through abuse by the injured party, the report will be of no further use for the purposes of litigation. It is up to the instructing solicitor to assess the quality of the report, whether or not it is fairly based on the evidence and whether or not it is worthwhile instructing another expert. Experts may differ as to the cause of the accident, especially where critical information is missing.
Experts can also be wrong. It is a sad fact that many investigators fail to probe deeply when presented with instructions either from within their own company, for internal reporting, or from lawyers. I have seen numerous examples of expert reports where the investigators have made a superficial inspection of the failed product, and drawn the wrong conclusions. Such reports are usually from one to three pages in length, and often include no plans or photographs of the product to support the conclusions.
The danger from the superficial approach to problems is that the whole legal process is put into motion on an unsupported base. The process continues to run, sometimes for years, until another expert produces a more firmly based report, which often contradicts the original. This builds up costs that might have been saved. The moral is that failure reports must from the outset examine the evidence in sufficient depth to draw reasonable conclusions.
So, how do forensic engineers approach failure problems? What methods are used, and what are the objectives?
The aim of all investigations is to determine what actually happened during an incident or accident, based on the material evidence. There is usually direct evidence in the form of witness statements, if litigation has already started, or descriptions from personnel involved. The failed object or objects must be inspected and examined, an objective that is usually achieved by photography. In addition, the investigator may wish to examine the scene of the incident personally to corroborate statements from others. The examination of the material evidence will be considered here, but as later case studies will show, documentary evidence is usually important for building up a comprehensive picture of events.
The methods described in this section are mainly, but not exclusively, used for examination of failed polymeric materials or products. However, some methods are of such generality they are used for any failed product, irrespective of the material of construction. Both macro-photography and micro-photography are therefore of universal application. On the other hand, chroma-tography is only of use in analysing polymers, because they are long-chain molecules. Metals and ceramics do not have such structures, so the method is inapplicable.
Many commercial disputes involve not just one failed product but several, in which case a representative sample, if not all the samples, need to be examined. Large numbers of failed products present their own problems, most important is determining how representative the sample is of the larger number alleged to be faulty (Box 8). Single failures present different problems. A sample, whether one of many or the only one, is amenable to the analytical tools that are now available in many laboratories. Care is needed to ensure as little as possible of that sample is affected by analysis.
There are broadly two ways of examining evidence: inductively or deductively. Inductive arguments rely on examination of a representative sample of failed products so that any conclusions are firmly based on the evidence available.
I was recently involved in litigation that alleged a medical product had been poorly made, did not conform to the dimensional requirements specified in an engineering drawing, and apparently did not function properly. The component involved was a transparent sight tube for breathing apparatus (Figure 18). The engineering drawing specified polysulphone plastic for the moulding.
Although over 100 had been made, the several different moulding companies had experienced problems with the tool supplied to them, and blamed it for their problems in trying to make acceptable mouldings. An independent expert appeared to suggest the tool was faulty, but that it could be rectified at extra cost. The manufacturer of the medical apparatus sued the toolmaker claiming the tool was faulty.
Another independent expert was engaged by the solicitor for the defendant toolmaker, and naturally wanted to examine the mouldings directly. It proved difficult, but after many months delay a few were passed over for inspection. They showed mainly moulding defects, and possibly some damage that could be attributed to the tool itself. Later, the expert was able to view a large sample of around 100 mouldings not released initially. A number were taken at random from the boxes in which they were packed. Inspection showed that the largest batch – made by one moulder – showed variable moulding conditions indicating they had problems with handling the polymer. It is well known that polysulphone is a difficult polymer to mould easily, for example, a hot tool heated to more than 100° C must be used. Initially, the moulders used water cooling and could not mould at all, but then managed to fit an oil unit for hot moulding. The sample of the large batch showed that all of the defects were caused by variable moulding conditions (Figure 19), with the exception of some marks from a blunt tool such as a screwdriver levering against the precision tool surfaces (Figure 20). This examination of a single piece of evidence is an example of deductive reasoning.
The expert for the claimant refused to consider the sample mouldings, and repeated his original view. The case collapsed after the claimant was cross-examined in the witness box on the third day of the trial.
Photography at the failure site is often invaluable immediately after a failure, either because the sample is too large to be easily removed, or because the extra evidence is crucial to the investigation. However, this is not always available, and the investigator must frequently rely on reports from loss adjusters or others who are first on the scene, before removal of the key material evidence.
The product itself usually reveals more about the failure than is apparent from a superficial inspection. If the product is paired with another, the second product can also be of value when examined closely. This is what happened when a failed crutch and an intact crutch were examined, as Box 9 reveals. Traces of wear on the pair of crutches corroborated the statement of the injured person, and helped to build up a picture of the circumstances surrounding the accident.
Macro-photography of the failed product is always the first task, before any disassembly or further handling of the sample. The aim is simply to make a permanent record of the sample in the as-received state, even if the features shown on the specimen make no sense. Subsequent evaluation will aim to explain all unusual or unexpected features.
An elderly woman suffered trauma and injury when one of the pair of crutches she was using to support herself suddenly broke at the junction of the aluminium handle and the main shaft. She was recovering from an operation to remove the lower part of one of her legs, and fell onto her stump.
Each crutch consists of an aluminium tube to which is fitted another aluminium telescopic arm to enable their height to be adjusted for different users. The crutch in question had fractured at the junction of the two tubes, where they were connected by a plastic insert (arrowed in Figure 21). The fracture surface on the insert was not especially revealing, but did show brittle-like behaviour, which was unusual for the material, a polypropylene copolymer.
Careful inspection of the interior of the polymer insert, with a stereomicroscope, showed that there were several subcritical cracks or crazes close to the edge of the fracture surface. They showed that the material had been loaded to exceed the tensile strength of the material. There was no evidence for excessive loads, and the material showed unusual discolouration inside the insert, but lower down than the subcritical cracks. The latter are incomplete cracks separate from the critical crack path. Such discolouration indicated some degradation had occurred and the part was thus defective. The wear patterns on the aluminium tubes confirmed that the failed crutch had been more heavily used than could be accounted for in supporting the woman (Figure 22).
After receipt of the report on the defective crutch, the injured woman received compensation for the trauma and injury.
Many interesting features can be found on failed samples by inspection with a hand magnifier. The features can be photographed with close-up attachments for a single-lens-reflex camera, such as extension tubes and bellows. However, detailed inspection of many features demands the use of a simple optical microscope, such as a binocular microscope with a magnification up to about × 60. The microscope can show scratches, abrasion, hairline cracks, crazes and contamination debris that might have been missed and discounted. If the sample is fractured, a detailed map from the microscope of the fracture surface is helpful in determining the sequence of events during crack propagation.
When surveyed, the fracture surface can show many specific features that indicate how crack growth occurred. Features such as the nature and direction of crack growth, and the age of the fracture can all be studied using a technique known as fractography.
Figure 23 shows the fracture surface of a polyacetal pipe fitting, which caused a flood when it finally parted.
The surface had clearly been formed a considerable time before water escaped, judging by the deposits that were present on most of the fracture surface. A fresh fracture showed where the sample finally parted when it rolled off the desk of the insurance investigator's desk after it had been removed from the accident site!
There were at least two zones of different colouration of deposit, which turned out simply to be inorganic salts from the local hard water supply (Figure 24a).
However, the deposit could not be removed to examine the underlying fracture in the polymer surface itself. This is a frequent problem with unique failed samples: several investigators may need to examine the surface independently, so any inspection must be carried out non-destructively.
However, subsidiary features can be used to work out where the origin, or origins, of the failure occurred on the surface. Where the pipe fitting was in service, several cracks must have started independently, because remnants of crack growth were detected at the outer edge of the fracture. The remnants consisted of flaps of material left behind after growth and caused by crack branching into the interior of the material. They indicated crack growth direction, so by surveying the whole edge, it was possible to infer there must have been several separate cracks. The flaps pointed in different directions, so must have been formed by separate and independent growth, enabling a more detailed map of the features to be constructed (Figure 24b). The cracks had grown from the base of a screw thread, a geometric feature that is a well-known stress concentration. Small weld lines were present here, so the combination allowed small applied stresses to have been magnified many times until the strength of the polymer had been exceeded. This is a common situation found in failed samples, where different defects combine together to produce failure. Small weld lines on the inner bore also probably grew to merge with the outer cracks. As you will see later, however, an additional factor was also present that produced an early fracture, but a late failure, in this particular sample.
Sometimes it is useful to carry the investigation further using scanning electron microscopy. This method uses an electron beam instead of a beam of visible light, and magnifies features that have molecular dimensions.
In addition, direct chemical analysis of any impurities, for example, is possible from the X-ray spectrum emitted by the sample. Box 10 relates how the method was used to solve a problem encountered on the Hong Kong underground railway.
The makers of the large lead-acid storage batteries used to power the underground locomotives wanted an independent investigation of a fire on the then new Hong Kong underground railway. The base of one of the storage batteries showed a large hole that was surrounded by burnt polypropylene (Figure 25). A close-up showed that a brittle crack bisected the hole (Figure 26).
Acid had leaked from the crack and caused considerable damage to the metal container. The shape of the hole seemed regular as though caused by impact with a hard object, as the SEM image of Figure 27 shows. SEM analysis of the edge of the hole showed the presence of aluminium and silicon elements, as well as the lead and sulphur expected from the inner plates of the battery (Figure 28).
The X-ray spectrum allows particular elements to be analysed from the peak position. The height of each peak reflects the concentration of that element present. There were substantial traces of aluminium and silicon present compared with slight traces of lead and sulphur. The traces were probably caused by an aluminium alloy stud present in the base of the container when the battery was inserted (Figure 29). The stud should not have been there at all. When the base of the battery impacted the stud, the impact damage initiated a crack, which then penetrated the interior of the cell, allowing acid to leak out slowly. It reacted with the aluminium and removed the evidence of the impact, apart from the traces left on the edges of the hole. The subsequent fire damage was not the fault of the battery manufacturer but the installer, who should have checked the floor of the holder was clear of debris before inserting the battery.
A common problem investigators face is determining the chemical constitution of the failed sample. A product specification is often absent, so an analysis route must be used. Whether there is a specification or not, the need for analysis is three-fold.
It establishes the nature of the material, and perhaps some of its properties.
It is a check in cases where a specification is available.
If the results show something odd for the material it could indicate material degradation.
Once the material is known, the engineer can form an opinion on whether it was the correct choice for that component. Property tabulations can be consulted for guidance.
To give a simple example, if a component is stressed in service, then it should be tough rather than brittle. Polystyrene is brittle so should not be used for stressed products – such as a crutch. A preferred alternative would be ABS or HIPS plastic, which are much tougher and crack-resistant.
If the material is exposed to a particular environment, then it should be capable of resisting that environment.
A fuel pipe should resist petrol or the fuel being contained, and rubber should resist oil if used for engine seals, water if used for a cooling system.
So what techniques are available? The technique chosen depends first on the nature of the material of construction, whether it be a metal, ceramic glass or polymer.
The first method used for polymeric materials is infrared spectroscopy (FTIR). It is a method that is partially destructive, so it is not used unless permission has first been sought and obtained. However, damage to the failed part can be minimised by selecting an area well away from key features because only milligram quantities are needed. Box 11 describes the use of spectroscopy as a forensic tool.
Polymers are characterised by their repeat unit, or units if copolymers. The repeat unit is simply the combination of atoms linked together, which is repeated in a linked linear chain. This repeat unit – a molecular block – has distinct properties that can be detected using spectroscopy, which works on the basis of absorption of electromagnetic radiation: infrared rays for FTIR, and ultraviolet rays for UV spectroscopy.
The sample from the failed crutch produced the spectrum in Figure 30. From that, it was possible to determine the material of construction as polypropylene copolymer. The group that distinguishes the copolymer as polypropylene is labelled on the diagram. The unknown polymer can be identified by comparing its spectrum with a printed compilation of spectra.
The comparison against a reference spectrum can also lead to detection of polymer degradation, as in the spectrum in Figure 31. It shows oxidation of a polyethylene material due to overheating during processing, probably coupled with attack by UV radiation from the sun.
Ultraviolet spectroscopy is another useful method for polymer analysis. The method is especially sensitive to aromatic groups present in a material. These are any structures with benzene rings (C6H6), and are common in UV absorbers and anti-oxidants. A material can be checked for the presence of these protective additives using UV spectroscopy.
Nuclear magnetic resonance (NMR) spectroscopy is another special tool useful in detecting minute amounts of impurities or special additives screened by other compounds.
The method described in Box 11 is not only used for polymers, but also for any organic substance. So it also finds wide use in forensic science for the analysis of drugs and textiles, and any other trace organics at a crime scene.
Spectroscopy is also used on commercial polymers to detect traces of impurities or additives.
Differential scanning calorimetry (DSC) is a tool used to corroborate polymer composition, and it can be used for low-melting-point metals and alloys. It is based on an entirely different principle to spectroscopy. It yields information on the thermal behaviour of materials – specifically the melting point denoted by Tm and the glass transition temperature denoted by Tg. These two pieces of data help characterise the polymer under test. The method is described in Box 12. The comparison of melting temperatures and transition temperatures with the standard values from published data will also show if the values are lower than expected for a high molecular mass material.
A method quite independent of spectroscopy is differential scanning calorimetry (DSC), where the absorption or liberation of heat is measured accurately as a few milligrammes are heated up at a regular rate. If the polymer is crystalline, heat will be absorbed as the melting point Tm is approached, with a sharp peak at Tm itself. The glass transition temperature T is observed as a change in slope of the curve. In addition, the degree of crystallisation will be shown by the area under the melting curve. So, if crystallisation was due to exposure of the sample to high temperature, for example, the melting curve will increase in size. This is what was observed with the elastomeric radiator seals in Figure 32.
An increase in the crystallinity of a polymer usually gives a stiffer material, explaining the greater stiffness of the failed seals. No change was detected in the melting point, which might have been expected with any degradation in molecular mass. However, stress corrosion cracking limits degradation to a small amount of polymer at the crack tip, therefore the change would not normally be found when examining bulk material.
The last method determines the molecular mass of polymers. One special problem with polymers is the drastic fall in strength with decreasing molecular mass, which often turns tough behaviour into brittle behaviour. If the part is safety-critical, it may not be able to resist normal stresses – let alone abnormal stresses – and will fail.
The method of choice for finding the molecular mass is GPC. It is a method based on absorption and desorption of chain molecules from a special gel. It gives a distribution curve for a typical polymer, from which various molecular masses can be calculated. GPC is used for:
checking molecular mass against the part specification – to establish the grade of polymer;
determining whether degradation has occurred.
Comparison of a new sample with the failed sample will show whether degradation – lowering of molecular masses – has occurred, and it can pinpoint the cause.
Comparison of samples taken from failed battery cases, for example, enabled ultraviolet degradation to be identified as the cause of failure (Box 13).
A common failure mode of polymers exposed to sunlight is degradation from the UV component. It can also happen unexpectedly for other reasons, when the product is normally concealed from sunlight.
Polypropylene (PP) is commonly used in battery cases that are usually hidden from light under the bonnet of a vehicle. Fork-lift trucks used by the Israeli army had lead-acid batteries that leaked after exposure to the sunlight of the Middle East (Figure 33).
Gel permeation chromatography was used to analyse the top and bottom surfaces from the degraded lids, with the results shown in Figure 34. The polymer from the upper surface possessed a much lower molecular mass than that from the lower surface. FTIR spectra showed carbonyl groups in the chains, a normal accompaniment of UV degradation, corroborating the initial assessment of the problem.
It was curious that the most severe degradation was restricted to the welded part of the lids, as can be seen from the chalking of the left-hand lids in Figure 33. The lids are welded to the boxes by heating the contact areas, and then clamping them together. Clearly, some thermal degradation had occurred, making the welded regions more susceptible to UV attack.
The manufacturer accepted responsibility for the problem, replaced the degraded batteries, and modified the material with a UV-absorbent additive to protect it against sunlight. The weld operation was also closely examined and modified to prevent unacceptably high temperatures.
A plastic garden chair made from polypropylene collapsed and injured the user after being outside for several years. The collapse was caused by cracking at an external corner, and there was evidence of extensive superficial cracking.
Construct a fault-tree analysis diagram for the fault. Suggest appropriate analytical methods for determining the most probable cause or causes.
The main possible causes of the cracking are mechanical overload, the presence of a stress raiser, UV degradation, and a moulding defect. Mechanical overload could mean the weight of the user exceeded the specification maximum load, or it could have been caused by impact damage. The stress raiser could be because the design of the corner made it too sharp, or perhaps wear had created a deep scratch. Moulding defects could include voids or contamination near the corner. My diagram is shown in Figure 35.
The polymer is known, so that material determination is not needed. However, the fact it has been exposed outside suggests UV degradation could be the prime cause of the superficial cracking. FTIR should be used first to see if any oxidation has occurred in the outer layers. Comparison with unexposed material – perhaps from the underside of the chair – could indicate the importance of UV degradation, especially if combined with GPC analysis of the two samples.
The critical crack should be examined with an optical microscope to see if there is any connection with the superficial cracks. If a superficial crack was found at the origin on the fracture surface, it is evidence that the UV degradation created the conditions for failure. The FTIR spectrum might indicate whether or not a UV inhibitor had been added to the original polymer. Microscopic examination should also discover any moulding defects such as a void or contamination near the corner, especially if the feature ran into the fracture surface.
An artificial composite leg fractured suddenly when a woman put her weight on it, and she was injured.
The composite was made by injection moulding a thermoplastic polymer – the matrix – reinforced with short glass fibres. The fibres added stiffness to the product and should have be aligned as parallel as possible to the long axis of the prosthesis.
Construct a fault-tree analysis diagram for the fault. Suggest appropriate analytical methods for determining the most probable cause or causes.
There are several possible causes of the failure, including many of those discussed in the previous SAQ, such as mechanical overload, and stress raisers on the surface of the leg. A composite material increases the possibilities because there are two different materials present: the fibre and the resin matrix. In addition, poor fibre distribution during moulding must also be included in the analysis.
My diagram (Figure 36) shows faults for the matrix material only and does not include fibre faults.
It would be appropriate to start by analysing the materials of construction: the polymer matrix and the fibres. FTIR is appropriate for both kinds of material, and a check using standard spectra should reveal the nature of the materials, and whether any degradation is present. GPC may be needed to confirm that the molecular mass of the matrix is appropriate.
The fracture should be examined with an optical microscope to find the nature of the failure, and to determine the origin, or origins. Careful inspection should reveal if the fracture is old and has grown slowly with time. Contamination on the fracture surface is one good indicator of an early initiation event or events. SEM might help if any contamination is found on the fracture surface.
The woman's weight may need to be compared with any specification; if her weight was greater than specification, the failure could simply be mechanical overload. The details of the sequence of events prior to the accident needs to be known to check subcritical damage – for example from an impact – so that all possible causes can be checked in detail. Elimination of the most obvious causes should then lead to the actual cause of the accident.
A battery case leaked acid from a crack in the outer casing, which was injection moulded. A nearby metal pipe leaked brake fluid and led to an accident when the brakes failed suddenly. There was evidence of corrosion on the brake pipe. There were also signs of impact damage near the site of the crack.
Draw a fault-tree analysis diagram to show the connection between the observed failure and the root cause or causes. Suggest the best techniques to pinpoint the problem. Indicate any uncertainties, and what extra knowledge might be needed to find the exact cause or causes.
Figure 37 is one attempt to outline the causal links between the accident and the problems visible in the engine compartment.
It is tempting to jump to the conclusion that the battery leak caused corrosion of the brake pipe. However, it is necessary to establish the chain of causation. For example, is the brake pipe close enough to the battery leak for acid to have made contact with the metal?
Analytical SEM inspection of the corrosion on the brake pipe could resolve the question, because battery acid contains sulphur, which can be detected by ED AX analysis. If it were not found, there is a chance the corrosion was caused by some other agent, such as acid rainwater or salt water.
Another obvious question to ask is whether the corroded pipe was situated vertically below the end of the crack? If so, there seems no problem in showing that acid dripped down onto the pipe. If not, further questions might be posed. Has the pipe been moved since the accident? Can a simple reconstruction show its original position? After all, a pipe in one position for some time will probably leave trace evidence of its position from oil and dirt always present in an engine compartment.
The damage to the battery that caused the crack, and hence the leak, needs careful investigation. Microscopy of the damage might show that the cratering corresponds to a hammer head. If so, the damage was almost certainly deliberate – perhaps to adjust the battery position in its recess.
Why did a brittle crack grow? Was there a stress raiser behind the impact blow? Is the material of sufficient quality to withstand impact damage? Is a hammer blow a normal and expected force that a battery case should be able to withstand? Were there any moulding defects found near or at the crack origin on the fracture surface?
Mechanical failure is a common failure mode for a wide variety of products. It therefore makes sense to conduct mechanical tests on the product to indicate the weak points in its design. It is important the product is tested under identical, or near identical, conditions to those expected in service. So, the load or stress system to which the product will be exposed should be analysed closely.
A common set of different types of stress is shown in Table 5, together with a set of typical examples. Most are simple, such as torsion, which mainly puts the sample under shear. But a bending stress is a combination that places the convex surface in tension and the concave surface in compression. In bending, there may also be a shear component if the section is thick enough.
Choosing the material of construction is an important part of the design phase. Suppliers usually make a material available in sheet form, from which samples can be taken for simple tensile testing to give values for strength and stiffness. What such tests do not reveal however, is the sensitivity of many materials to the effects of product geometry.
There are many reasons why materials in products fall short of the strength predicted from the simple tensile test. One of the most important reasons is the way applied stresses can be concentrated at particular points in products. Such points are known as stress concentrators, and their effects can be predicted using relatively simple mechanical models.
Figure 38 visualises the stress lines in loaded samples. Owing to the restrictions of product geometry, the lines are squeezed past the tips of such features as cracks. The net result is that a stress concentration exists at the tip of a crack. Sharp corners in products are the source of similar stress concentration.
Perhaps less easy to appreciate are the effects of internal voids or internal cut-aways, but these too can cause serious stress concentrations, as Box 14 describes.
Stress concentrations can be modelled mathematically by several different methods. The stress concentration at a notch tip in an elastic material can be modelled relatively easily. The stress, sigma (symbol σ), at the notch tip, is given by the equation
where σ0 is the uniform stress applied to the material sample; D is the depth of the notch from a free surface; and r is the radius of curvature at the notch tip.
In the centre of Figure 39 are two examples where the above equation can be used. At the left and right of Figure 39, there are other features that will concentrate stress. If there was a cylindrical hole through a flat bar the stress would be three times greater because D = r (two semi-circular notches side-by-side). A spherical void would produce a stress two times greater because the stress is distributed over the larger internal surface area of the void.
The design engineer can use standard reference works to estimate the effect of other stress raisers. A textbook example is shown in Figure 40, where a sketch and a graph represent a notch in a plate subject to bending. Figure 40 is versatile because it can be used for a variety of design configurations. Finite element analysis programs are also available, and are based on standard formulae.
It is possible to model particular configurations physically. The method involves constructing a model in birefringent plastic such as polyurethane or polycarbonate. When it is strained under the expected conditions and polarised light is shone through it, a fringe pattern shows the stress concentrations.
You need to be aware of escalation when stress concentrations interact. The net effect is the product of the individual factors, so it is more serious than expected. Thus, a sharp crack at a notch tip is more serious than either alone. Similarly, a circular groove inside another circular groove produces a net effect of nine times greater because the net effect of the two stress raisers is the product of the effect of each individual stress concentration.
Stress concentrations are ameliorated in many tough materials because yielding occurs at the critical zone, especially if the sample is loaded slowly. However, it is not always possible to predict the effect, so stress raisers due to shape should be ameliorated wherever possible. Where the material is known to be brittle, or where fatigue is possible – even in a tough material – every attempt should be made at the design stage to reduce stress raisers.
The effect is normally described in terms of the stress concentration factor Kt. This is simply the ratio of the nominal imposed stress σ0 and the actual stress σ at the feature in question
For any Kt value greater than unity, the actual stress at the stress concentrator will be greater than the nominal imposed stress, and if the actual stress exceeds the strength of the material, the sample will fail. If a brittle crack is initiated at a stress raiser, the net section area becomes smaller. The crack therefore accelerates until the stress is relieved. Again, depending on sample geometry, it could result in the product breaking into two, or possibly more, parts with a consequent loss of function.
What points in the solid thermoplastic products in Figure 41 are most vulnerable to mechanical failure by cracking? Estimate the stress concentrations at those points under the loading condition shown by the grey arrows. For (a) r = 0.25 mm, t = 2.5 mm; for (b) r = 0.4 mm, t = 8 mm; for (c) r = 0.1 mm, t = 2 mm, D 0 = 22 mm, D 1 = 30 mm.
Use Box 14 as a guide. How would you reduce the weakness in the products?
In Figure 41(a), the weakest part of the product is the internal corner. The stress concentration can be estimated by using Figure 40, in Box 14, with the given dimensions.
You can use the diagonal distance between the two corners for d and then you can calculate r/d. Pythagoras' theorem or trigonometry will give you d.
Reading from the curve of Figure 40 gives
So an impact on the corner will cause a stress about 3.25 times the nominal applied stress.
In Figure 41(b), an impact against the flat side of the ribbed plate will produce a bending load on the rib corners, so they represent the weakest parts of the structure. Using the same method as above
r/d = 0.4/11.3 = 0.035
Reading from Figure 40 gives
In Figure 41(c), the thread root is the weakest part of the system, and the part will fracture from here when a bending load is applied. The stress concentration can be estimated as before using Figure 40.
The distance from the root to the inner surface is
r/d = 0.1/2 = 0.05
Reading from Figure 40 gives
Kt = 3.75
Figure 42 shows a simple box made from rigid plastic filled with liquid. Assume when dropped, it will suffer impact on one of the corners. Knowing the radius of curvature r at the inner corners is 0.1 mm, and the wall t is 4 mm thick, estimate the stress concentration at the centre of each corner, and as a result, predict which corner is most likely to fail. What would be the effect if the radius were to be increased to 1 mm? What other steps could be taken to improve product performance? Refer to Box 14 for help with the problem.
The stresses can be evaluated using Figure 40, Box 14. For the right-hand corner
r/d = 0.1/5.66 = 0.018
Reading from Figure 40 for an r/d value of 0.02 gives
So this corner will experience a stress concentration of about six times the applied stress.
The situation for the left-hand corner is more complicated, because there is a spherical void near the inner corner. The net stress raising effect will be the product of each stress raiser: the inner corner and the void. Using the multiplier 2 from Box 14
So failure will probably occur at the left-hand corner.
The effect of increasing the corner radius ten times to 1 mm can simply be evaluated in the same way, using an r/d value of 0.18. Reading from Figure 40
Kt = 2.2
This represents the stress concentration for the right-hand corner; that for the opposite corner is twice this, giving Kt = 4.4. The relative situation between the two corners is unchanged, although the severity is much reduced, by about a factor of three. Note that a further increase of radius would produce diminishing returns as the curve flattens out at higher r/d values.
An alternative design strategy might involve changing the depth of material at the corner by either increasing the overall wall thickness or thickening the corners locally. The origin of the void may be a result of poor quality material or poor manufacturing, so attention here could solve that problem.
With incidents where a product or process failure is suspected, a recollection of events often helps. A recollection brings together the known facts, and puts them in a logical sequence. It is a useful, if theoretical, exercise because putting events into a chronological sequence can throw up unexpected problems, which must then be answered. If the existing hypothesis cannot explain all the known facts, a fresh hypothesis may need to be constructed. However, a practical or experimental reconstruction can be even more valuable if it throws up new and unsuspected features of an incident. This is what happened in the incident considered in this case.
The case study in this section deals with a ladder accident and a practical reconstruction of the circumstances. Ladder accidents are common both on industrial and in domestic premises. Official statistics indicate it is one of the most common causes of sudden death in accidents in the home, especially among elderly men.
The accident occurred on a dry, summer morning, just after the home owner had erected a two-stage extension ladder, which had only been bought recently. He put it against the back wall of his house to clean the windows on the first floor of a two-storey detached house. The ladder was standing on the level concrete paving slabs of the patio adjacent to the back garden of the house.
According to the witness statement, the owner was at or near the top where he was washing the top window (Figure 43(a)). As he moved down the ladder to clean the pane below, the ladder suddenly slipped from the sill and ‘walked down’ the wall (Figure 14(b)). He fell and was seriously injured. The ladder ended up at right angles to the wall. One of the plastic tips at the top of the ladder's uprights – called stiles – had broken (Figures 44(a), (b) and (c)), but the feet were intact. There was no damage to the metal ladder or any visible defects.
It was thought the broken tip could have caused the fall, by allowing the ladder to slip down the wall. So the owner of the ladder, who was also the injured party, initiated action against the ladder manufacturer. The solicitor approached an expert for a report, following funding from the Legal Aid Board.
The key evidence was the broken tip (Figures 44(a) and 44(b)). It showed a brittle fracture in an unknown, relatively rigid plastic. The fracture surface was fresh, and there was no sign of old or subcritical cracks. FTIR analysis showed the material was in fact a copolymer of polypropylene, which should normally be tough and ductile. But what did the fracture reveal, and were there any obvious features that could represent defects exposed on the surface?
Sketch the fracture surface from the photograph (Figure 44(b)) and isolate the main features on the fracture surface. (Tracing paper can help with this.) Include any features adjacent to the surface that could be defects involved in the fracture event. Attempt to interpret the features on the fracture surface itself, given that a fracture origin is often surrounded by a half-moon shape that shows where an initiation crack starts to accelerate during its growth. Relate your origin to the external features using Figure 44(c), and indicate whether the external features could cause a concentration of stress. Indicate the severity of the stress concentration, and therefore account for the fracture.
Figure 45 shows the fracture surface of the broken ladder tip with main features indicated.
The void at the centre of the surface is a serious stress raiser, and probably makes the Kt value about 2. Because it is symmetrically positioned in the section, stress from any direction is likely to have the same effect.
There is a line nearby on the outer surface, but it would have a small effect. More serious is the sharp external corner.
The central void is clearly a stress concentration. However, analysis of the crack surface itself showed the origin to lie elsewhere. It was a moulding parting line at an external corner produced by the tool used to injection mould the component. It is where two mating parts of the steel tool meet, and it often shows a mismatch owing to tool wear (Figure 44b). The conjunction of this small corner and the larger external corner thus produced a larger stress concentration than elsewhere on the corner. Although only one of the two tips had broken, both showed severe abrasive damage to the parts in direct contact with the wall just before the accident, as just seen in Figure 44c.
Draw an FTA diagram for the possible causes of the ladder accident – including the broken tip and any other possible causes. Sketch a simple diagram of the situation to help find the causes of the accident. Consider the condition of the ladder, the state of the ground on which it was leant, the angle of repose and degree of extension of the ladder. With your collection of data, indicate the most likely causes of the accident.
Figure 46 shows the simple situation of the ladder leaning against a wall, with points of contact with the ground and wall.
Figure 47 indicates some of the main possible causes of the accident.
The hypothesis most favoured was that the tip broke and allowed the ladder to slip off the sill, unbalancing the user, who fell off. In turn, the tip may have broken because of faulty design – and perhaps faulty moulding – or by an impact blow caused by movement of the ladder by the user. The scenario is shown by the causation path in bold. However, other possibilities include ladder instability from wet ground or from too low an angle of repose. The failure of metal parts was not supported by examination of the ladder, so was excluded.
At this stage of the investigation the causes of the broken tip could include:
There was little support for poor material from either DSC or spectroscopic analysis (Figures 48 and 49). In Figure 49 is the result obtained from the feet of the ladder, which appeared rather more flexible. They were probably made from a polypropylene with more ethylene repeat units present. This would make the material more elastomeric, with a lower level of crystallinity. No carbonyl groups were present at the characteristic wavelength (Figure 48); so the material could not have been degraded in processing, for example. Although the void represented poor design and moulding practice, it did not initiate the crack that broke the tip. The sharp corner in the moulding had initiated the crack and it might have been a design defect if it had failed and caused the accident. Sharp internal corners are common in many moulded products, but are not necessarily defects unless it can be shown they cause product failure.
Given the lack of positive evidence for the cause of failure, it was then essential to consider the conditions under which the ladder was being used and the mechanics involved. A ladder leaning against a wall is an application of simple static analysis to determine the critical variables that can affect stability (Box 15). The key variables are the angle of repose of the ladder, and the coefficient of friction between the feet and the ground.
The ladder problem is a classic in applied mechanics where it is important to know the safe angle at which to lean a ladder against a wall (Figure 50). This angle – the angle of repose, labelled θ – is made by the ladder with the ground. The ladder is at equilibrium, so the forces in the system are balanced. However, if the system of forces is changed by someone climbing to the top, the safe angle is changed. A change in frictional forces would also be critical.
The forces in Figure 50 include the reaction of the ladder (length x) against the ground and wall (N, R) and the weight of the ladder (Mg). When the equilibrium equation is simplified, N and R can be eliminated, giving
where μ0 is the coefficient of friction of the ladder feet against the ground, and μ1 is the coefficient of friction of the ladder tips against the wall. The static coefficient of friction being considered here is the value when all surfaces are completely still – the coefficient becomes smaller when the surfaces start to move.
If a rather low value of 0.25 is assumed for the coefficients of feet and tips, the angle of repose of the ladder at equilibrium is 62°. This means the ladder will slip down at lower angles, but is stable at angles above this critical value. What happens if the user climbs to near the top? The equilibrium equation tends to
The critical angle rises to 76°, so the ladder is less stable. Any angle below 76° results in instability, when the ladder can slip away. Note the exact position of the user on the ladder is not specified in the equation owing to simplification.
What about the value of the friction? In fact, values of the static coefficient of friction are usually higher than 0.25, a typical value for a wooden foot against a wooden floor being about 0.5. This gives a lower critical angle of repose, and makes the ladder less likely to slip. A wet floor has low friction, so must be avoided. It is recommended that the repose angle be about 75°, and for extra security someone stands on the lowest rung, or the ladder is tied to the wall at the top or the base.
A visit to the site of the accident was essential. The witness statement could be checked, and any further evidence that could clarify the circumstances could be examined directly. The visit showed the visible evidence of the intermittent contact between the tips of the ladder and the wall below the window (Figure 43(a)) and lower down the traces of an impact with the small sill above the patio doors (Figure 51). The trace contact marks corroborated the witness statement, showing how the tips and hence the ladder structure itself, had oscillated as it slipped, giving intermittent contact with the wall. No traces of marks from the feet on the concrete slabs of the patio itself could be found. The key information about the stability of the ladder, the angle of repose and coefficient of friction of the feet, remained unknown, however.
The injured user attempted to guess the information requested – especially the angle of repose and degree of extension of the ladder – but it is always best to determine such information independently. After traumatic injury, someone is more likely to forget details, even if in normal circumstances they could recall them.
The ladder had not been photographed in situ just after the accident. The lack of trace marks from the feet meant the extended length of the ladder could not be determined. However, there was a key bit of information that seemed indisputable: the tips of the ladder were resting on the window sill, which was 3.69 metres vertically from the ground. This was measured during the site visit.
The events were reconstructed using the ladder fitted with new tips. As a starting point, the ladder was fully closed rather than being extended. Another ladder of different design was placed alongside for the purposes of comparison and safety during the stability tests. The objectives of the test were as follows.
To estimate the static coefficient of friction of the feet.
To establish the angle of repose.
To determine the footprint of the feet.
To load the ladder statically at the estimated height of use.
To examine dynamics by simulation of movement.
The first task was to estimate the static coefficient of friction of the polypropylene feet. A simple way of doing this for very light loads is to place the feet on a surface similar to that being used for the real ladder and tilt it until the feet just slip (Figure 52). In this case, the surface is a concrete slab. Theory (similar to that in Box 15) shows the critical condition for slip on the inclined plane is
where θ is the angle of inclination of the slab to the level ground. Two experiments established that the coefficient of friction for the feet – mass about 100 gm – was approximately 1.0.
Such a value is typical for the elastomeric polymer concerned acting on smooth concrete. But did the feet fitted to a ladder give the same value in the reconstruction?
Direct determination from the unladen accident ladder showed that it slipped at an angle of about 40°; so, neglecting friction at the tips, the coefficient of friction is about 0.6, considerably lower than the value estimated from an inclined plane experiment. There are several reasons for this situation, as Box 16 relates.
A classical law taught to generations of engineers states that the static coefficient of friction – friction force divided by normal force – is independent of load and contact area, and it is roughly true for most rigid metals and ceramics. Polymers, rubbers, fibres, thermoplastics and thermosets, composites and all natural materials, such as wood, bone, and skin are quite different and deviate substantially from the classical law. The effect has been well studied, as Figure 53 shows for an important speciality polymer, PTFE.
The effect has interesting consequences. For example, it means fibrous polymers hold together well in textiles where a high coefficient of friction acts between the fibres. PTFE fibres in GoreTex waterproof matting are stable and won't pull apart easily. On the other hand, PTFE sheet, with a low coefficient of friction of about 0.04 under load, will ease the movement of heavy machinery where the sheet is used as a bearing pad.
Elastomers generally show high frictional coefficients (greater than 1) because most imposed loads will strain all of the material, and not just that at the contact zone. This is a unique and desirable property, and is used in products like tyres, which must grip the road under variable road conditions.
In the first place, the coefficient of friction of polymers is known to be dependent on load, decreasing as the load increases. Secondly, the feet of the accident ladder had been designed for a repose angle of about 75° (Figure 54), any other angle reducing the contact surface against the ground. This was demonstrated by recording the footprint of the feet at several angles of repose (Figure 55). The edge in contact with the ground thus becomes even more heavily loaded over a much smaller area of contact.
The next issue to be addressed was what angle the ladder had been leaning at. The claimant stated that it had been leant against the sill of the upper window. So the angle of repose could be calculated for two situations: an unextended ladder and the ladder extended by one rung. Greater degrees of extension would create progressively lower angles of repose. As a working hypothesis, it was assumed the ladder had been used either unextended, or with one rung extended. Leaning the ladder against the sill would produce a repose angle given by
The length of the unextended ladder was 4.46 m, and extended by one rung it was 4.71 m. With the sill being 3.69 m from the level ground, then θ = 56° unextended, or 52° for an extension of one rung. The situation for the unextended ladder is shown in Figures 56(a) and 56(b), with the plastic tips leaning against the window sill. In this position, a slight movement of the ladder would allow the ladder to jump down onto the adjacent wall, where either the aluminium tips of the lower section of the ladder, or the plastic tips on the upper section would make contact with the brick wall.
The tips from the accident showed abrasion, most visibly obvious at their upper corners (Figure 57). In addition, matching the ladder to the wall with known lengths and heights showed the unextended ladder had most likely been used. The angle of repose of 56° was well below the recommended angle of repose of 75° for this design of ladder. But what could explain the curious set of marks below the window and the comment from the user about ‘walking down the wall’?
The final phase of experiments involved simulating the weight of the user when working near the top of the ladder. Both the weight and height of the user were known, and he thought he was standing on the fourth, fifth, or sixth rung from the top when the ladder slipped. The user could thus be simulated by simply suspending a fixed mass of 72.6 kg (force of about 726 newtons) to represent his weight on a rope from the upper rungs. The mass of the ladder was about 20 kg (force of 200 N). It was leant at an angle of 56° against the wall, representing the unextended ladder (Figure 58).
The exact position of the suspended mass was important, because when ascending or descending a ladder, the user would have shifted his weight from foot to foot. The user could not remember which rung he was standing on at the time of the accident, but it was likely that the fourth, fifth and sixth rungs from the top were in use at the time. As the user was cleaning the right-hand pane of glass, his left hand was probably holding the left-hand part of the uppermost rung. By leaning over, his weight would have shifted to the right-hand part of the rung he was standing on, so the right-hand part of the fifth rung was used to hang the mass.
One feasible trigger for the accident could have been momentary loss of contact of the left-hand tip with the sill against which it was resting, by reaching over to the right, for example. A spring balance was used to determine how much force was needed to pull the left-hand stile at right angles from the sill – it was about 100 N. The torque at the top of the ladder simultaneously pulled the left-hand foot out of contact with the ground, and the loss of contact caused the right-hand tip to slip down to a slightly lower position against the wall. Repeating the effect led to progressively lower positions against the wall until the whole ladder became dangerously unstable and the experiment was halted.
What might be the expected effect of trying a stability experiment using a suspended weight from the sixth and fourth rungs? What would you expect to happen for each experiment? Using experimental values for the static coefficient of friction for various loading conditions, estimate the effect of increasing mass on the critical angle of repose for top-loading conditions.
It is easy to see the stability of the ladder decreases as the user ascends the ladder, so suspending the weight from the sixth step will make the ladder more stable than the fifth step. Conversely, suspension from the fourth step will make it less stable.
The stability equation for the situation of a top-loaded ladder is given approximately by
Adding a suspended weight to represent the mass of the user however, increases the total mass of the system acting at the ground, and hence the normal reaction. This will, in turn, increase the load on the polymer feet. The first experiment to measure the static coefficient of friction with an inclined plane gave a value of about 1.0 for a mass of 100 gm. But it was reduced to about 0.6 for the ladder fitted with the feet (20 kg). The total mass of the ladder plus suspended mass was 20 + 72.6 = 92.6 kg. For top loading (user near the top of the ladder), the critical angle θc will be described by the above equation, with the following estimates using the stability equation above.
The effect of the increased mass on the feet would be to lower the coefficient further, but by how much? Figure 53 suggests the biggest drop occurs at lower loads, the curve tending to flatten out as the load is increased further. So there is a smaller decrease as the load is increased further. Assuming that polypropylene behaves like PTFE, then a decrease of the coefficient of friction to about 0.5 might be expected. The critical angle of repose would be expected to rise even further to about 63°. Therefore, a ladder inclined to the ground at 56° would be unstable when top-loaded, and would slip. Moving the weight higher would decrease stability, while moving it lower would increase stability.
It was felt important to determine the limits of the stick-slip motion seen in the hung-mass experiment. There was no downward motion observed at all with the mass on the sixth step. However, the results for the fourth step were more dramatic.
Application of a force of about 120 N to the left-hand tip of the accident ladder led rapidly to uncontrollable stick-slip motion down the wall, and the ladder fell away (Figure 59). The marks left on a painted board visibly demonstrated the stick-slip motion, with a series of impacts showing the intermittent contact of the ladder tips down the wall (Figure 60).
What did the reconstruction show? Firstly, it suggested the ladder had been leant against the wall at a maximum angle of about 56°, well below the recommended angle of repose. Secondly, it confirmed that at relatively low angles of repose of this particular design of ladder, stick-slip motion could occur after momentary instability – even if the angle of repose was above the critical angle of repose. The instability was produced by a torque load that moved one tip of the ladder away from the wall. Provided the user was near the top of the ladder, catastrophic and uncontrollable loss of the ladder was inevitable.
The reconstruction of the accident showed the user initially leant the ladder at too low an angle for a reasonable safety margin. Although it was above the critical angle of repose, it was susceptible to stick-slip instability when the user was near the top of the ladder. The visible contact evidence from the wall (Figure 43(a) and (b)) confirmed stick-slip motion of the ladder tips.
But the question of the fractured tip remained unanswered. It could still have caused the accident if the tip had broken at the sill, and initiated stick-slip motion. However, both the broken and intact tips showed abrasion against brickwork, so it is more likely that it survived for some way down the wall. The final piece of evidence was an impact mark on the small wooden sill above the patio doors (Figure 51). This was probably caused when the tip hit the sill, and the tip broke from its weakest point. The fracture was a result of the accident, not the cause.
Construct flow diagrams for the separate sequences of events of the accident based on:
the initial evidence of the fractured tip;
all the available evidence.
Indicate what effect each sequence would produce on the tips of the ladder. Examine the actual condition of the tips to show which was the most likely sequence of events.
The sequence of events from initiation of the accident based on the initial evidence, assuming the tip broke to cause the accident, is as follows.
(a) Ladder tip fractures.
(b) Ladder slips from sill.
(c) Ladder starts to slip down wall.
(d) User thrown from ladder.
(e) Ladder hits ground.
The broken tip would not show any abrasion, because it would have been thrown clear after the first break, while the intact tip would be deeply abraded. Also, the sharp edge of the broken tip still on the stile might also be abraded by contact with the wall.
The sequence of events using all the available evidence is:
(a) Ladder tips slip from sill.
(b) Ladder tips impact wall.
(c) Repose angle drops.
(d) Ladder walks down wall with increasing velocity.
(e) Repose angle decreases rapidly.
(f) Sideways rocking motion of ladder throws user from ladder.
(g) Ladder tip hits sill above patio door and fractures.
(h) Ladder hits ground.
In this case, both the tips will show equal abrasion due to motion against the brick wall, as was found by inspection (Figure 57). The evidence points towards the tip breaking as a result of the accident, and not itself breaking to cause the accident.
Given this conclusion, the Legal Aid Board – which funds many cases brought by injured claimants – decided the case did not stand a good chance of success if it did proceed to trial. The victim of the accident did not therefore receive compensation.
The moral of this forensic story is that initial perceptions of failure may not always survive the scrutiny of probing analysis. The investigator should always examine the circumstances surrounding a product failure. The fracture was just one piece of evidence contributing to a reasonable interpretation based on all the evidence. Ladder accidents are common, and a site visit is often essential to gather more evidence of the incident. In this case, the contact evidence on the wall was important for the construction of a working hypotheses to explain the accident. In addition, practical reconstruction of the accident using the failed ladder demonstrated stick-slip motion above the critical angle of repose, and helped explain the contact evidence. However, the answer to the problem lay not in the tips but the feet of the ladder, because it is the feet that play the most important part in the stability of ladders.
After car accidents, accidents involving falls are one of the most common causes of death and serious injury in the UK, according to the Department of Trade and Industry. The DTI monitors accident statistics for one common cause of falls, those from ladders, and the figures show 42 per cent of accidents involved the victim falling from the ladder. The ladder broke or collapsed in 16 per cent of cases, and the ladder fell while the victim was on it in 22 per cent of cases. The victim tripped over the ladder, caught fingers on the ladder and so on, in the remaining 20 per cent of cases. Wooden and metal ladders were about equally represented.
What can be done to lower the accident rate? Informing users of the potential dangers of ladders and the best way of using them. Indeed, new ladders have one of the most elaborate warning notices posted on any product (Figure 61).
Owing to the instability of lean-to ladders, they should be stabilised wherever possible, by tying the feet or tips to a solid support, or by having a colleague standing on the feet to prevent slippage. Ladders should never be used when the floor is wet owing to the lowering of the coefficient of friction, and the floor should also be level so that both feet are in contact.
Ladders must be inspected at regular intervals, and the critical feet and tips renewed when worn or damaged, a service many ladder manufacturers actually provide for free.
But can ladder design be improved so as to lower the chances of slippage accidents of the kind considered here?
One of the problems of the accident ladder (Figure 54), is that it possessed fixed attitude feet, so that any deviation from a repose angle of 75° lowered the contact area with the ground. As the feet are the key to the stability of ladders, one solution is to provide feet that rotate as the angle of repose is changed (Figure 62). Then the full area of the feet is always in complete contact with the ground, so providing an extra margin of safety.
The design of the plastic tips and feet can also be improved for ladders of the specific design considered here. Although the tip broke during the accident rather than causing it, the design possesses several stress concentrators: one by design (sharp corners), the other from manufacture (internal void). Both could easily be eliminated, the first by simply smoothing the corners of the metal tools, the other by more careful control of the moulding process. Both procedures would increase the strength of the product substantially.
While the investigation did not support the claim against the manufacturer of the ladder, it did expose areas where this specific design could be improved, and also pointed to other designs of extension ladder that gave greater protection to the user. This is usually a common feature of many forensic investigations, simply because they probe not just the specific circumstances of particular accidents, but because they compare the design of products that either have failed, or appear to have failed at first glance, and suggest ways product design can be improved for future users. Product designers can therefore use the reports of such investigations directly to improve product performance. It can also aid bodies such as the Consumers Association in evaluating the effects of product design on accident statistics, so that the public can be made aware of the dangers and limitations of existing designs.
List the way the ladder investigation progressed, stating the various questions that the investigation threw up, and which the investigator posed as a natural outcome of his analysis of the failure. What were the major lessons concerning the investigative procedure? What role did the analysis of the statics of a ladder leaning against a wall have on the investigation? Indicate the importance of external literature on the interpretation of the experiments.
The investigation developed as follows.
User injured when ladder slips.
Broken tip found at accident site, so user infers that ladder accident may have been caused by product fracture.
Solicitor consulted by injured user of ladder.
Solicitor commissions investigation and report.
Investigator examines broken tip. (Questions thrown up: How did fracture occur? Where was origin? What features on the tip initiated the fracture?)
Fracture analysis shows defects in moulded tip, but corroboration of accident conditions needed. (Questions thrown up: What corroborative evidence is needed? What was angle of repose of ladder? How high was sill from ground? Was the ladder extended or not? Any other trace or contact evidence from accident site?)
Site visit to measure various distances and heights, inspect damage to wall, inspect for traces of contact of feet with ground, and so on.
Contact evidence seems to indicate ladder tip may have broken when it hit a lower sill, so further research needed.
Reconstruction of accident organised. (Questions thrown up: What was the coefficient of friction of the plastic feet? What was the critical angle of repose of the ladder? What was the weight of the user and on what rung was he standing when the accident occurred? How does the coefficient of friction of polymers change with loading conditions?)
Experiments show ladder slipped at low angle of repose. Stick-slip motion demonstrated, so explaining marks on brick wall at user's home. (Questions thrown up: How did the abrasion damage to tips of ladders occur? How does this evidence fit with existing theories of the accident?)
Re-examination of tips reveals both tips show abrasion damage, confirming tip broke after contact with wall rather than initiating accident.
Results indicate user did not use ladder under recommended conditions, so Legal Aid Board refuse to support further action.
Experiments also indicate improved designs for ladder feet, so that contact with ground constant at different repose angles.
In hindsight, the investigation showed the importance of corroboration. It was tempting to jump to the conclusion that because the tip fractured from the inherent defect of the void, the tip therefore caused the accident. The investigation showed the interpretation of one piece of evidence – the broken tip – is simply not enough to establish a case: all the evidence must be examined to construct a sequence of events. In this case, the evidence of abrasion to both the intact and broken tips could not support the initial hypothesis.
The examination of the stability of ladders was important in showing there was generally a critical angle of repose, above which the ladder was apparently stable, below which a ladder could slip uncontrollably. At or near the critical repose angle, instability could induce a stick-slip downward motion of the ladder. Experimental reconstruction of the accident showed exactly what angle of repose had been used, and how unstable the ladder was at this repose angle. Static analysis also showed the importance of the feet in providing a high coefficient of friction, and hence maintaining stability.
External literature showed that the coefficient of friction of polymers is not constant, as conventional mechanics suggests, but varies substantially and significantly with applied load. When a ladder has an adult standing on a rung, the coefficient drops, and the critical repose angle rises, making the ladder more unstable and hence increasing the risk of slippage.
The polymer tips and feet of the ladder considered in Section 4 are the most vulnerable to wear and tear because of their contact with the wall and ground. They are the safety-critical parts of the structure, not just for extension ladders but also for step ladders, loft ladders and other domestic and industrial ladders. So are the seals briefly mentioned in Section 2 of this unit. Such components are cheap and easy to replace, but the costs of either not doing so or using the wrong replacement can be serious. Like seals, containment materials must be resistant to the fluids they contain especially where the fluids can be hazardous. The two applications considered in this section and the next are a car radiator reservoir and a bulk storage tank.
The problem of the quality of materials and manufacturing processes used to shape those materials, apply to all systems in a car cooling system. A car radiator reservoir must resist the high temperature of the water,high pressure, and the high temperature in the engine compartment. New materials, such as glass-reinforced nylon, have become widespread for applications under the bonnet, and have proved themselves in applications such as inlet manifolds sitting directly on the engine. However, manufacturing standards must be high to eliminate defective mouldings.
Problems can be created by faulty design of storage tanks, where polymeric materials have good resistance to corrosion from powerful chemical liquids. The dangers of tank leakage, however, mean the design must meet minimum standards.
The storage tank case study raises the problem of standards in general, where national institutions often cannot supply appropriate guidance. It occurs because of the often inevitable delays in forming a committee to draft standards, so standards do not become available until several years after design and manufacture have begun. In the case study, a German standard was available to the engineers concerned, but was not apparently applied to the tanks in question.
In Section 2, you saw that high-temperature seals in central heating systems must at least resist hydrolytic degradation, and no less is true of new radiator materials. Radiator reservoirs must resist water temperatures of 70–80° C, high internal pressures and high external temperatures.
New materials such as glass-reinforced nylon, have become widespread for under-the-bonnet applications, and have proved ideal for inlet manifolds for example (Figure 63). The reasons for their adoption are multifold: they can be moulded into complex shapes so that air flow into the engine can be controlled effectively. This is a requirement of advanced fuel management systems.
Interestingly enough, inlet manifolds are made using metal cores that must be melted out after each injection moulding cycle, a new process route known as lost-metal moulding. This overcomes the so-called re-entrant angle problem faced by conventional injection moulding, where the core must be capable of being withdrawn at the end of the cycle to release the moulded product.
Such manifolds are lighter than cast metal equivalents, so that total engine weight is kept down and makes the car more efficient. A subsidiary bonus comes from greater ease of installation during assembly in the car factory. On the deficit side of the costs equation comes the greater capital cost of tool and machine investment than conventional materials, and the problems of developing new materials.
Such developments have encouraged use of, glass-filled nylon (GF nylon 6,6) for other car engine parts, and in the case examined here, car radiator reservoirs. The case study is presented in detail in Paper 3 which is attached as a pdf and should be printed out (if possible) to gain the maximum benefit from the discussion of this case study. The paper describes in a scientific format, the way the failure investigation developed, what initial inspection revealed and what tests were used to explore the failure. The results are discussed below by extracting key information needed for analysis, but you will need to read the paper in detail to gain a full understanding of the problem. As in succeeding case studies in this block, information not given in the original paper is also provided here. It will give you a wider perspective of the problem.
Click on the 'View document' link below to read Paper 3 (keep the paper open on your desktop for reference throughout Section 5).
A nylon radiator reservoir failed after a small number of journeys in a new model of car, causing engine seizure owing to the sudden loss of cooling water. The car had not been introduced into the market, and was being tested by the manufacturers. As in so many such investigations, the critical failed part was the focus of initial attention, and was inspected macroscopically first. It enables the investigator to stand back from the actual failure zone, and place the whole component in its context. This in itself can prompt key questions that may lead on to the solution to the problem.
The reservoir, which failed catastrophically, was fitted to a flat part of the engine, and comprised just a half-shell (Figures 1–3, in Paper 3). The reservoir was 41 cm long and 11 cm wide. Comparison of a failed product with an intact equivalent is always useful, and in this case showed the failed part was highly distorted (Figure 3, Paper 3). The centre part of the reservoir had contracted significantly when compared with a new reservoir. The same photograph also shows the way the part was made, by injection moulding from a central gate, so leaving a remnant sprue in the centre of the inside of the component. The sprue is circular and rather large because a wide gate is needed to allow the highly viscous molten polymer to penetrate all parts of the metal tool during moulding (Figures 3 and 6, Paper 3).
Why should the distortion be important? It clearly distinguishes the failed reservoir from a new reservoir, so it might be critical in determining the cause or causes of the failure. What then could cause such distortion? A factor common to all shaping processes is frozen-in stress or strain. When material is cooled from its molten state into the final product, such stresses and strains can be caused by over-fast removal of heat from the material. In polymers, the problem is one of frozen-in strain, and is discussed in greater detail in Box 17. Frozen-in strain can provide a driving force for crack growth, as well as causing distortion and therefore mismatch with mating parts. In this case, the distortion had probably been revealed by exposure to the hot water of the cooling system, at a temperature not far from the melting point of the material of about 265° C (see the DSC trace for the material shown in Figure 64).
Although there was no obvious connection between the distortion and the crack, it was an observation to be kept in mind as the investigation progressed.
Because polymers are poor conductors of heat, they require long cooling cycles during injection moulding. However, if the cooling rate is increased by chilling the tool, polymer chains can be frozen into an unstable state. The usual motive for chilling the tool is to lower the cycle time and hence increase return on the high capital investment of machine and tools. Although the product appears normal, any rise in temperature over periods of time can allow the chains to relax to a more stable state, producing distortion in the product. In addition, frozen-in strain can provide energy for crack growth.
The problem came to particular prominence with the development and application of so-called engineering plastics in the 1970s, when materials like polycarbonate were used in demanding safety products like miners’ lamps. The cases for the lead-acid batteries suspended from the miner's waist were found to have cracked, spilling acid onto the miner, and led to loss of charge. The situation was serious, with whole collieries out of action because no lamps were available for the work force. Investigation of the situation at one colliery showed lamps had an average life much less than the specification of three years before being withdrawn, as Figure 65 shows.
What was the cause? It was a combination of chill moulding by a supplier together with solvent welding – which initiated crazes, followed by cracks. And the design exhibited numerous sharp inner corners that lowered the impact strength substantially. Crack growth was driven by the frozen-in strain, and occurred quickly under the demanding working conditions in the collieries concerned.
The solution to the problem involved tool modifications to remove stress concentrations – by chamfering cores – and the use of hot oil circuits so that the polymer was cooled slowly during moulding. The failing material was filled, so a transparent grade was substituted. Unfilled grades are frequently stronger owing to the lack of stress-raising filler particles. The levels of frozen-in strain were reduced by the slow cooling, and a quick monitoring method enabled faulty mouldings to be identified easily and rejected. The method involved viewing the new transparent cases using polarised light (Figure 66).
The technique is completely non-destructive and easy to use once quality standards have been established. The modifications increased the life of lamps substantially (Figure 67).
The critical crack was found at one end of the reservoir, as shown by the arrows in Figures 2 and 4 in Paper 3. It was a single brittle crack that adjoined an external bracket used for attachment of other engine parts. The arrow on the left indicates the tide marks formed when the leakage occurred. As they ran in one particular direction from one end of the crack, it was inferred that the reservoir was oriented vertically with the lower end being that shown in Figure 2, Paper 3. The detailed view of the crack in Figure 4, Paper 3 is approximately life-size, so the crack was about 65 mm long, running lengthwise along the side of the reservoir, along a corner of an external buttress.
Inside the reservoir, the crack possessed the same shape and dimensions as externally, but ran from or into a different feature, known as a weld line (Figure 5, Paper 3). Weld lines in general are a regular and normal feature found in mouldings, normally where the molten polymer is forced around a core – to form a hole in the final product, for example. The unusual property of this weld line was that it fell nowhere near such a design detail. The fact the crack also lay very close provided a further clue to its origin.
Further information on the reasons for the weld line became evident when the material of the reservoir was analysed. Although detailed chemical analysis was not needed, because it was known from the product specification, it was important to see how the weld line had formed and whether there were any more structural irregularities. Several methods were used to examine the failed sample.
Dusting of exposed surfaces.
Sectioning, polishing and etching.
Dusting surfaces with a fine white powder called whiting could reveal surface features more clearly, and in fact such an elementary method did just that, as Figure 6, Paper 3 shows. The reason is that moulded composites often show surface roughness due to the orientation of the fibres. The picture shows several prominent flow lines – precursors to weld lines – denoted by the arrow at left, and a totally unexpected feature, a fragment of an original granule from the moulding process (arrowed at right in Figure 6, Paper 3).
Secondly, sectioning a new reservoir showed features – potential defects – inside the reservoir. Typical sections are shown in Figures 7 and 8, Paper 3. They have been polished and then etched to remove superficial debris and reveal the resistant glass fibre structure more clearly. Not only did the method reveal flow lines in the moulding, but it also showed up internal voids in the structure. However, as the method is necessarily partially destructive, it was not felt appropriate to apply it to the failure specimen, especially as there were other non-destructive methods available.
Because the material was a composite of short glass fibres and thermoplastic nylon, it was thought X-ray radiography could yield useful information on the pattern of fibre orientation in the failed sample. The method is well established as a non-destructive technique (Box 18) for many different materials, and so could be used directly on the failed specimen. It was useful in two ways, first, for showing internal clumping of fibres, seen rather dimly in the upper radiograph of Figure 9, Paper 3, and secondly, for showing the crack in outline (lower radiograph of Figure 9). Although medical, or soft, X-rays were used, the contrast in the radiographs was low, and the method turned out to be of limited benefit.
While both DSC and spectroscopic methods generally rely on partial destruction of the sample, there are other routes to determining chemical composition. A non-destructive test (NDT) does not destroy any of the critical sample, so it would be favoured because it allows further examination. An example of such a technique was given in Box 17, it involved the use of polarised light to reveal frozen-in strain. Most macroscopic and optical microscopic inspection methods are also non-destructive, provided the sample is not dissected for viewing.
There is another reason why failed samples should be preserved intact as far as is possible: where litigation is started, other experts will want to examine the specimen independently.
Many methods have been developed over the years, especially with fatigue sensitive materials such as aluminium alloys in air frames. Some can also be used on polymeric materials. Soft X-rays are useful for showing the internal structure of composite materials, and even thermoplastics. They also reveal internal cracks.
Soft X-rays were used in a dispute concerning gas injection moulding, a new technology used to make the arms of office chairs. A contract was based on several sample arms moulded with modified tools at the contractor's factory abroad. Unfortunately, the technology proved inadequate for making strong arms. In particular, the exact position of the internal gas pocket could not be controlled accurately and reproducibly. The arm needed to be very strong where it was attached to the chair shell, but often the gas pocket penetrated this region, weakening the product drastically. As the trial approached, the original demonstration samples were discovered and produced in evidence. The defendants would not allow the samples to be sectioned physically, so the expert for the claimants suggested a trip to the local hospital to radiograph the arms non-destructively. One such radiograph is shown in Figure 68 where it can be seen that the gas pocket – the light grey – has penetrated the attachment zone where the two moulded holes are.
The other arms showed a similar problem. A key metal part of the machine that operated the gas injection phase had also failed by fatigue (and been mended twice, on the instructions of the defendants) indicating a serious problem with the method. The dispute went close to trial. The judge visited the factory to see the machine for himself, but the evidence of faulty arms was overwhelming, and the dispute was settled just before trial.
The settlement was a good example of a fair agreement that satisfied both parties. The defendants had many new conventional machines, which could not be sold owing to depression in the market, and the moulder needed to re-equip his factory. The deal involved the moulder taking enough machines at a discount equivalent to the damages he was requesting, so goodwill between the companies was preserved. The technology of gas injection moulding has advanced, but the problems of appropriate design and reproducible manufacture should not be forgotten.
Of greater use where fracture surfaces are concerned is microscopy, initially at low magnification with a simple optical microscope, followed by scanning electron microscopy. It is sensible to be reluctant to deliberately break a unique specimen, but as all other methods had been exhausted, valuable evidence should be shown by the fracture surfaces. This turned out to be the case, in fact.
Figure 10 in Paper 3 presents the optical micrographs of one side of the fracture surface, which at first sight seemed to show rather little in the way of a clear origin or other features that can often enable a picture of crack development to be reconstructed. Careful inspection however, revealed tide marks left on the outer buttress. As there were several such features, it was reasonable to suggest several cracks had existed before the final failure. The evidence for that event lay in a much larger tide mark below, showing cooling water debris had collected there following a major loss of water.
The second interesting feature seen at low magnification was a line of smooth material on the inner side of the fracture. This could represent unfused polymer present within the weld line already detected. Such indeed proved to be the case when the sample was examined in the SEM (Figure 11, Paper 3). Owing to the greater resolution of the method, vague or ill defined features from the optical microscope can be seen with greater clarity. The top picture already shows substantial detail in the fracture, with a three-dimensional view unequalled by optical microscopy. The sequence of shots is of increasing magnification from (a) to (c) as shown by the captions. Shots at greater magnification were taken, but revealed nothing new that couldn't be seen at lower magnification. The resolution is in fact so good that individual fibres can be seen. The picture at (c) shows the flat region at the base, or inner side of the fracture, to be quite smooth with no fibres.
One hidden bonus of SEM, at least in this case, came from the need to coat the sample with gold to give a conducting surface, which is essential with non-metals to bleed away electrons from the main beam. When the sample was re-examined optically, the contrast and definition were much improved, as the panoramic sequence of Figure 12, Paper 3 shows. The weld line surface was now shown to be running for a large distance along the fracture surface. But more significantly, there were visible numerous cold slugs of partly melted granules embedded in the surface – shown by the longer arrows in the photograph. Their presence seemed to confirm a problem of cold moulding in the sample. The problem occurs when the moulding machine is started up ready for production. The barrel heaters are still warming up, so the polymer granules are not fully melted and homogenised before injection into the tool.
What about the mechanical strength of the material? It was important to test the material directly, and compare new and failed polymer samples. Tensile testing would provide basic mechanical properties, which could, for example, be used for comparison with the specification provided by the material supplier.
In a material showing orientation, either of fibres or the polymer molecules themselves, it is an obvious ploy to test the samples both parallel and at right angles to any flow marks. Such tests could reveal any anisotropy and fundamental flaws such as weld lines, and so corroborate independent results from microscopy. The raw load-elongation curves are shown in Figure 69, and the results of analysis are on page 192 of Paper 3. They show the best results fall below the ideal suggested by a specification from the material supplier – about 80 MN m−2 compared with 140 MN m−2. In one case, the tensile test showed the sample to fall well below the ideal value, when tested laterally to the flow marks in the sample. The material appeared rather brittle, although of high modulus. The results therefore seemed to confirm the material could be weakened seriously by flow or weld lines present. Could mechanical analysis throw any light on the reason or reasons for failure?
List the features that could be defects within, and on, the structure of the failed radiator reservoir. Draw an FTA diagram to describe the likely causes of the failure, and indicate the most probable cause of failure, stating the evidence for your opinion. In hindsight, what were the most useful experimental methods used in the analysis described in Paper 3?
The following represent features that could be defects within the structure, and on all the visible surfaces of the failed reservoir:
weld and/or flow lines visible on the inner surface;
cold slugs both on the inner surface and inside the material, as exposed in the fracture surface;
sharp external corner on the outer buttress;
voids within the structure of the reservoir;
change in fibre density or orientation.
Figure 70 shows the possible causes of failure. Low initial water levels, lack of topping up or other unsupported theories have been omitted.
The most probable cause or causes of failure must explain the premature failure of the radiator under apparently normal driving conditions. The evidence of the tidemarks on the corner of the buttress suggests progressive failure from small cracks that grew in the wall of the structure. In the absence of any known external stresses, it is most likely that such cracks grew under the influence of the internal pressure acting on the most serious defects in the composite structure of the wall. Although the material showed poorer mechanical properties than the specification from the manufacturer, it seemed less likely than defects within the structure concentrating stress to unacceptable levels.
The basic reason is that new reservoirs did not show the same level of defects such as weld and flow lines, cold slugs and so on. Defects in the material could clearly lower the strength of the material (sample No. 4 on page 192 of Paper 3) .In addition, there appeared to be a high level of frozen-in strain when the failed and new reservoirs were compared (Figure 3, Paper 3). The evidence thus appears to point to either faulty design or manufacture, or perhaps a combination of both possible causes.
The most useful methods of analysis were simple comparison of new and failed reservoirs (to show distortion and flow and weld lines); microscopy of the fracture surface and immediate surroundings (to show the defects within the material); and tensile testing to show the variation in strength due to defects.
What stresses is a radiator reservoir subjected to in service? They can be calculated from the internal pressure of the system using standard formulae (Box 19). The water in the cooling system is under pressure, and one value suggested by the manufacturers was 25 psi. This is equivalent to a pressure of about twice atmospheric pressure, or 0.1725 MN m−2(1 psi = 6.9 kN m−2). When values of r = 22.5 mm and t = 2.5 mm are inserted in Equation (7), Box 19, it produces a hoop stress of about 1.55 MN m−2, a value well below the measured strength of the worst sample of about 55 MN m−2. What could cause the material to fail? In the discussion in Paper 3 it is suggested that four kinds of stress raisers that could initiate a brittle crack (page 193).
Geometric stress raiser at the buttress corner.
Cold slug fragments.
Estimating the net effect of all these factors working together is difficult. Using the known dimensions of the buttress corner, it was possible to use a stress concentration diagram (Figure 40, Box 14) to estimate a Kt value of about 4.2. The model chosen to represent a void at a cold slug, for example, was that for a penny-shaped crack (Figure 71). However, the evidence for the existence of such a shape in the fracture surface was rather weak, because distinguishing a void from a crack growth region was difficult from the micrographs. Nevertheless, a value of Kt of about 6 emerged from the analysis, so that a net stress concentration of about 25 (6 × 4.2) could have been working to weaken the wall near the buttress corner.
The final factor was the distortion in the reservoir noted earlier. It was possible to calculate the effect it could have produced from a tensile curve made during mechanical testing, together with the observation of the widening of the crack in the whole failed sample reservoir. The analysis gave the surprisingly large value of about 20 MN m−2 (page 194, Paper 3). Thus, the net effect on the potential initiator was for a total stress of about 59 MN m 2. This value is now comparable with the strength of the weakest sample tested, so it seemed reasonable to conclude that the combination of defects initiated one, or more likely, a series of cracks, which created the final leak.
A pressurised pipe can be analysed relatively easily for the stresses on the pipe wall, and this is surprisingly useful in a range of practical problems. There are two stresses at work in a thin-walled pipe: the hoop stress σH around the circumference, and the longitudinal stress σL acting along the main axis of the pipe (Figure 72). (A thin-walled pipe is defined as a pipe where the wall of thickness t, is less than about a tenth of the radius r, so t is small compared to r. For simplicity, consider there to be an internal pressurep that is constant at all places.)
Analysis shows the hoop stress is the largest stress in the wall and is given by the equation
The longitudinal stress is half the hoop stress and is given by the equation:
When the wall is more than a tenth of the radius, separate equations must be used, although they have a similar form to the above equations. An important inference from the above analysis is that longitudinal cracks grow preferentially to hoop or circumferential cracks. As cracks grow at right angles to the load, longitudinal cracks grow from the hoop stress, while hoop cracks grow from the longitudinal load on a pipe. If hoop cracks are found in a pipe, it therefore implies they have grown under a different load regime than simple hydrostatic pressure.
Draw a diagram to show the most likely sequence of events that ultimately led to the failure of the reservoir. Indicate on the diagram the evidence on which the sequence is based. Show the original cause of the failure, and discuss how the manufacturer of the radiator assembly could tackle the problem with the moulder of the reservoir.
Based on the comprehensive analysis of potential and actual defects found in the fracture surface, it is possible to construct the sequence of events in Figure 73. To the right is shown the evidence that pointed to each event.
The cause of failure was faulty moulding of the radiator reservoir. The most likely origin being when an early moulding was placed by the operator in the tub intended for delivery to the assembler of the parts intended for the new car. As several cars were being tested, and only one radiator failed, it is unlikely the other reservoirs were poorly moulded. The overall moulding procedure was probably correct, the fault lying with the operator who mis-identified the sample.
The problem could be tackled by informing the moulder of the cause of the problem, and asking for a review of quality control procedures. This could include both discussion of instructions given to operators and sampling procedures used after moulding.
The results of the investigation were clear-cut. The problem was caused by a combination of defects in the material of the wall of the reservoir. They, in turn, had been mainly caused by faulty moulding. One scenario by which they were produced has already been suggested: the failed reservoir was made during startup of the injection moulding machine, when moulding conditions had not been established. The failed reservoir was therefore a single maverick. Several conclusions could be drawn.
In the first place, the radiator manufacturer could be reassured that the basic design was not at fault, and they could not expect to see widespread failures on new cars. Secondly, they could ask the supplier to re-examine the quality policy. It is normal and good practice for moulders to supply each operator on the machines with a diagram of likely defects to look for in mouldings as they are made. This is the first line of defence in quality control, and perhaps the most important, because every component is – or should be – examined individually.
The faulty reservoir must have been made at this stage, and had been mistakenly accepted as a good reservoir, and forwarded through to the manufacturer. The moulder should be asked to check that such quality checks were indeed in place, so that the buyer might be assured of products meeting the specification. The moulder generally has responsibility to supply quality products, but it is also true that development designs made in short runs are not always examined as closely as is necessary.
So, several useful and practical suggestions to improve the situation were thrown up by the investigation. Inspection could be improved at the audit stage in quality assurance, by lightly dusting with whiting to show up any surface defects, such as cold slugs and weld lines, which would show whether there was a problem in moulding. The design could be improved by increasing the radius of the buttress corner. Sharp radii are easy to ameliorate by simply polishing the corners of the core that produce the corner in the tool. The effects are often dramatic, especially when the material of construction is brittle, as in this case.
By interpolating on the appropriate stress concentration diagram, suggest a specification to provide the toolmaker when correcting the buttress corner radius of the radiator reservoir. By examining the relevant photographs in Paper 3, suggest what other areas of the moulding should be inspected by the toolmaker for possible improvements. Could any further improvements in the material be recommended, either as a development project or for routine design development?
One appropriate stress concentration diagram to use is Figure 40, Box 14. The wall thickness of the reservoir is 2.5 mm, and the radius of curvature at the buttress root is about 0.1 mm (Paper 3, page 193). The object of the exercise is to reduce the stress concentration at the butress root.
To reduce the stress concentration to unity is impossible within the limits of the diagram. Moreover, the diagram suggests the goal would be unrealistic because the curve falls only very slowly to unity as r/d increases. If the far right-hand edge with a value of r/d=0.4 is used, the value of r is
r = 0.4 × 3.5 = 1.4 mm
which gives a stress concentration of about 1.6. Such a value of Kt is substantially below the values of 4 or 5 assumed above, and could be acceptable to the design team – provided that good mouldings are available. The change to the tool would simply involve polishing the sharp edge to the core that creates the feature on the final product.
There are several other similarly sharp corners on the mouldings shown in Figures 1–3, Paper 3. They include the buttress at the centre of the reservoir (Figure 2, Paper 3), the projecting bosses shown at either end of the reservoir in Figure 1, a corner bracket (also in Figure 1 left) and the inlet and outlet pipes (Figure 1).
The material choice could also be reviewed with the material supplier to find a grade of improved strength without sacrificing stiffness or any other desirable properties. There is wide experience of the material type in under-the-bonnet applications such as inlet manifolds, so it is unlikely a shift to an entirely new material would be feasible. This in itself would require new trials, and it is by no means certain the original tool would be capable of moulding in a new material with different flow properties.
The case study has been considered here in some depth because it provides some useful lessons in forensic methods. Composite materials of the kind used in the reservoir are of relatively recent widespread application, so ways of studying them are still being developed. This is why the author of the paper included many methods that were not especially critical in the final analysis of this particular problem. However, the point is that they could find application in other failure cases of this kind of material. Other investigators can follow up further details of the methods using the references supplied at the end of the paper.
One surprising outcome of the investigation was the utility of the gold coating method for increasing the resolution of optical microscopy, rather than the original intention of aiding SEM. Nowadays, gold coating would not be used because non-conducting samples can be examined directly with ESEM (environmental SEM). This is a new development that will be examined later in this block, but samples do not need coating, so the method is completely non-destructive in nature.
Other developments with SEM, especially in image recording and sample stages, mean that very large samples can be examined directly in the microscope. Large samples need not be cut down simply to fit into the microscope, a common problem with older SEM machines. It was also surprising how the simple and relatively cheap method of dusting with whiting, revealed surface detail not normally visible.
Another interesting result came from tensile testing: the moulded material was substantially weaker than the supplier's specification. It is true with many other polymers, perhaps because suppliers frequently either choose best samples, or choose the best results – subconsciously or not – to promote their product. Designers should be aware of the problems of taking and using supplied data without allowing for discrepancies.
The second containment case study deals with storage tanks. They are products that must be correctly designed to perform their function of containment of a hazardous fluid safely over a reasonable lifetime. The consequences of failure can be catastrophic if the stored contents are released into the environment, and so it was in the example described in Papers 4 and 5. It was a tank made by the thermal welding of sheet polypropylene, and was designed to store 30 tonnes of concentrated caustic soda. Papers 4 and 5 are attached as pdf documents which should be printed out (if possible) to gain the maximum benefit from the discussion of this case study (they should at least be kept open on the desktop throughout Section 6).
Click on the 'View document' link below to read Paper 4.
Click on the 'View document' link below to read Paper 5.
A brief description of the hazards of caustic soda is useful at this point (Figure 74). It is one of the most corrosive agents in common commercial application, being a by product of the electrolytic manufacture of chlorine. Figure 74 shows the hazard notice of the fluid caustic soda, and warns of the dangers. It is a syrupy liquid with a concentration of about 40 per cent caustic soda (NaOH). It is extremely aggressive in contact with a wide range of materials, especially living tissue. Human skin, for example, is rapidly dissolved with no pain to the victim because the nerves are consumed at the same rate as surrounding tissue – unlike attack by strong acids. Escape into streams and rivers would kill a range of plant and animal life.
Plastic tanks have been in use in industry for a long period of time, at least from the early 1970s, and continue to expand in usage owing to the ease of fabrication, the excellent resistance to corrosive chemicals, and the low cost of the material. The tanks are made individually to order by hand construction with fairly simple equipment, so manufacturing costs are low.
The increased usage encouraged smaller companies to enter the field with competitive prices. But the only standard covering their design, use and testing is German, and not easily available in English in the UK – although partial translations can be obtained from the larger manufacturers.
This case study is based on an investigation carried out after catastrophic failure of a virtually new tank in Warrington, Cheshire. It had been installed at the premises of a small business manufacturing ‘dairy detergent’ (Paper 4). The latter is a dilute solution of caustic soda to which various other chemicals are added to aid the cleansing effect; one use is for flushing out dairy equipment on farms. The 40 per cent caustic soda was the basic raw material for the final product, and just four deliveries had been made to the factory when the failure occurred. Each delivery involved complete filling of the tank.
The failure fortunately occurred after the shift had left for the day, and was spotted by the production director, who was alone in the office above the factory floor. He heard a bang, followed by what sounded like rain, and when he looked out onto the factory floor, he saw a jet of fluid shooting across the open space (Figure 1, Paper 4). The protective wall around the tank completely failed to contain or constrict the fluid, which ran along the floor and into an adjacent unit that printed tachograph discs. Here the damage was extensive, because specialist printing equipment was attacked and wrecked. The local fire brigade were called on to contain the fluid, and prevent escape into the outer environment. Eventually, a specialist recovery company was hired to remove the fluid.
On inspection, the evidence of failure was slight. The tank was intact, with only a small gap in the middle of a welded seam in the centre of the panel to show a serious accident had occurred (Figure 3, Paper 4). The crack had been opened up – deliberately by other investigators – so that it traversed almost the entire panel, when inspected several days after the accident. The crack was at the dead centre of the weld, where the panels had been joined together thermally.
Draw an FTA diagram to show the possible causes of failure of the tank, starting, as usual, with the accident event itself. Include any other possible hypotheses in addition to those mentioned above. Indicate broadly what tests might be needed to clarify the problem of isolating the actual failure mechanism. What other information could prove useful in the investigation?
Figure 75 shows some of the main possible causes of the failure. As the failure occurred early in the product life, wear or other long-term mechanisms are not included. Weld quality is clearly an immediate candidate for inclusion in the diagram, but other failure mechanisms such as material attack, or choice of material, should be included.
The kind of analytical tests needed are:
material quality checks using FTIR and DSC (check welds and check bulk material);
mechanical testing of weld and bulk integrity;
macroscopy and microscopy of the failed surface.
Other information that could be useful includes a material specification, and details of the way the tank was actually constructed. A specification for the tank itself is probably the most critical piece of information. Such a safety-critical product would almost certainly require detailed design calculations of wall strength and stiffness to ensure a given lifetime. Whether or not there are applicable standards would also need investigation, probably starting with the British Standards Institution and the American Society for Testing Materials.
It was clearly vital to examine the fracture surface, being apparently the only piece of forensic material evidence available about the incident. It had to be removed using a circular saw (a rather lengthy procedure with such a large tank, although cutting the soft polymer was like slicing cheese). The key fracture surface is shown in Figure 5, Paper 4. The fracture surface was simple, and showed:
four distinct zones;
a clear origin;
vertical flutes across all the surface.
It is easy to suggest the boundary to each zone represents a period of slow crack growth following each complete fill of the tank. Notice also, how the size of each zone increased with each fill, a feature to be expected from a growing crack. The origin (O1) appeared to be a small, elongated pit, or pin-hole, in the outer surface of the panel.
Use Figure 76 to evaluate the magnitude of the stress concentration at the pin hole void at O1. Use information from Figures 5 and 7, Paper 4 to give you a basis for quantitative assessment of the severity of the defect. What can you say about the stress system acting on the tank? Use your ideas about the stress system to explain the shape of the cracks as they grew slowly under the load. There were other subcritical cracks (see Figure 6, Paper 4) that had grown from even larger pin-holes at other parts of the weld in the critical weld. Explain why they did not grow to criticality.
Figure 5, Paper 4 shows the fracture surface at the origin of the critical crack that led directly to the leak in the tank. The dimensions of the original pinhole can be found from the photograph. The wall thickness is 12 mm, as stated in the caption to the figure. In the blown-up photograph, the wall is about 44 mm, therefore the photograph is about 3.7 (44/12) times bigger than life size. So all measured dimensions must be reduced by this factor to give actual sizes. The pinhole dimensions are:
depth from free surface is 1.62 mm (6/3.7)
width along fracture surface is 0.81 mm (3/3.7)
The lateral diameter of the defect can be estimated in a similar way from Figure 7, Paper 4, assuming that the subcritical pinholes shown in an alternate weld are representative of the critical weld defect. The width of the weld was actually 2.5 mm, so the photograph is magnified by 4.8 (12/2.5) – it is about 5 times bigger than life size. The pin hole in the centre of the crack near the centre of the plate is about 0.5 mm as measured directly, so is actually about 0.125 mm (0.5/4.8) in lateral diameter.
These dimensions can now be used to estimate the stress concentration. Figure 76 shows an elliptical hole of major axis 2a and minor axis 2b. Using the dimensions calculated from the photographs
a/b = 0.8/0.125 = 6.4
Reading from Figure 76 using this value shows the stress concentration factor will be
However, it is worth bearing in mind the curve rises rapidly as a/b increases, so there is uncertainty about the value of Ktg. Also, the lateral diameter of the hole was estimated from an adjacent weld, and could be unrepresentative.
The tensile stress is applied at right angles to the long axis 2a of the hole, which in the case of the tank, will be a hoop stress in the wall. It derives from the hydrostatic force from the contents of the tank. There will be some variation through the thickness of the wall, the maximum stress occurring on the outer surface, which is why the cracks have grown to a greater extent along the outer surface when compared to their inward growth.
The crack grew with successive loading cycles, showing the increased stress concentration with increasing crack size, that between propagation zones 3 and 4 in Figure 5, Paper 4 being the largest, and ultimately critical stage of growth. A crack from zone 4 penetrated through the wall therefore allowing an opening in the weld to release the caustic soda in a jet on to the factory floor.
The subcritical cracks shown in Figure 6, Paper 4 occurred off-centre, where the imposed stresses were lower. The load was greatest at the centre of the vertical weld, which explains why a smaller defect there was critical while larger pinholes elsewhere in the weld never became critical.
Similar comments apply to Figure 7, Paper 4 and it can be inferred that the central crack in Figure 7 was less serious in its stress concentrating effects than the critical crack. The stress will have been roughly the same as in the adjacent panel, but the cracks would have grown at a lower rate.
So the picture that emerged from the fracture surface was of a weld failure. But did this mean the weld itself was faulty, was the material sub-standard, or had the material been attacked by the caustic soda? The design of the tank might also be at fault.
Further inspection of the tank was needed to check the other welds. Those welds at the same level would have been subject to the same hydrostatic pressure from the caustic soda.
The reason for examining those welds is that the pressure in the tank at the time of failure is given by the equation
where p is the hydrostatic pressure of the contents, ρ the density of the fluid, h the height from the top fluid surface and g the acceleration due to gravity. Pressure therefore varies with height, as shown in Figure 77, for fluids of varying density. In this case, as the density is fixed, and g is a constant, the only factor controlling pressure is the vertical distance from the surface of the liquid to the weld height. If a vertical weld had failed at a particular height, there could be evidence of other cracks at nearby welds at the same height.
The hunch proved correct, but only in one of the four possible welds (W4 as shown in Figure 1, Paper 4). The weld was obscured by its position, making inspection difficult (looking for a dark crack on black material in the dark); although when fine talc was used to dust the weld, it revealed several subcritical cracks (Figure 7, Paper 4).
But what did it indicate? It might show the welding was faulty, but did not explain the uncracked welds (W2 and W3). Were there any other clues to the cause? Another factor could be overloading: was there any distortion visible in the tank? After all, the distortion visible in the radiator reservoir described in the previous case study was a significant clue to that problem.
If Figure 3, Paper 4 is re-examined, such distortion does indeed exist. It is best seen by tipping the page and viewing the edges of the far side and the crack at an angle. Both features can be seen to be bowed outwards, the cracked weld showing the greater effect. The effect was confirmed by direct measurement of the circumference at several heights: a difference of 20 mm in a total of 8.55 m being measured for the centre of the cracked weld compared with a higher panel. Although the effect was small, it indicated possible overloading, producing creep of the panel material.
The quality of the welds and the sheet material also needed to be assessed independently. Samples from both intact and cracked welds were sectioned, and the sections polished and etched to show any internal structure. No serious problem with weld structure could be found (Figure 8, Paper 4). The tensile strength of the welds also proved reasonably consistent with one another and the manufacturer's figures (Table 1, Paper 4). Chemical analysis of the weld and bulk materials using DSC proved negative, the melting points all being close to one another. FTIR showed no obvious anomalies, such as an oxidation peak.
Although weld strengths appeared reasonably high, there was a small difference of about 4 per cent between the tensile strength of welds from intact and cracked panels. Could mechanical analysis be used to check the stress concentrating effect of the pinholes? The first point to check was the formula needed to calculate the hoop stress in the vessel. Hoop stress acts around the circumference of a tank, or pressure vessel.
Fortunately, the stress situation in a static pressure vessel is simple. Where the ratio of tank radius to wall thickness is greater than about 10, the tank can be treated as a thin-walled pressure vessel. The ratio is in fact
and the hoop stress, σH is given by Equation (7), Box 19.
Knowing r and t, and calculating p using Equation (9) with a known density, the hoop stress at the failure crack was evaluated as
It is now possible to calculate the stress concentration that would have been acting at the critical pinhole, shown in Figure 5, Paper 4, prior to the first growth of the crack. Because from Equation (3)
From the mechanical tests of welds, it was possible to calculate that the real stress acting at the pinhole in the centre of the lower single-thickness panel was about six times the hoop stress from the hydrostatic pressure of the contents. A cross check on the figure is possible from SAQ 20, where Kt was estimated at about 14 times the nominal applied stress. As mentioned in the SAQ, there is some uncertainty in the assumed diameter of the hole, which could cause a large error in the estimate.
So, there appeared to be no serious problems with the material, processing or fabrication, but could the design of the tank be queried? The structure is shown in section in Figure 4, Paper 4, comprised essentially of 12 mm single thickness panels, which have been buttressed by three extra hoops of material at the base, centre and the top of the structure. But is this the best way of resisting hydrostatic load?
Equation (9) shows a simple linear relation between hydrostatic pressure and height for a given fluid (Figure 77). If that is the case, should not the thickness increase gradually with distance from the top? A dam increases in thickness from top to base to resist the water pressure, and so the same principle should apply to any fluid reservoir.
By adding hoops, the top hoop could be redundant, and the lower hoops might not be sufficient to resist the greater pressures towards the base of the structure. Such an hypothesis would explain why the failure occurred in a lower, unreinforced panel. It would not explain why only two welds failed from pinholes, but the calculation above shows the single thickness panel is having to resist a large hoop stress. Doubling the wall thickness would halve the applied stress, while having three panels here would give a stress of only about 1.15 MNm−2. Even if pinholes occurred in the weld, failure would be much less likely with such substantial lowering of the hoop stress.
So, the general conclusion of the stress analysis was that the design itself was faulty. To resist hydrostatic pressure, the tank should have been designed like a dam, rather than like a barrel (Figure 10, Paper 4).
Why should just two of the four welds have shown cracks? The answer came when the welding stage was inspected directly. The hoop of panels for such tanks is made sequentially by hot fusion welding, that is, by melting the surfaces of two panels and pushing them together. This is fine for three of the welds where the flat panels are joined, but difficult for the final joint when the ends have to be brought together by bending the sheet into a cylinder (Figure 9, Paper 4). It would certainly explain the lower quality of one weld, and the poor quality of another weld was probably caused by similar problems in bringing two large flat sheets together. The quality of such welds is tested for through-the-thickness holes using a spark tester, a method that will not detect partial pinholes. One rather disturbing aspect of the process is that the hoop so formed is under a bending stress, so the outer surface of the final wall will be in tension. This will, of course, make failure more likely, and is akin to the frozen-in strain problem of the radiator reservoir described previously. In this case, there is a frozen-in stress rather than strain.
It was possible to calculate the effect from simple bending theory, as described in Section 5, Paper 5 (page 222). The effect probably added about 1.5 MN m−2 to the hoop stress, so that the above calculation under-estimated the stress. This changes the effect of the stress concentration, reducing it to about 3, enhancing the effect of less serious pinholes.
The investigators concluded that the design of the tank was faulty, having been made like a barrel rather than a dam. Failure was inevitable, and explains why failure occurred so early in the tank's life. Because the walls were exposed to excessive stresses, it was inevitable the tank would fail quickly. The stress would seek out the weakest welds in the most exposed panel, and failure was not caused by faulty welding at all. That is not to say the welding process could not be improved, for example, by annealing panels and carefully controlling bending before final welding.
But the failure should never have occurred in the first place because there was a standard for such thermoplastic tanks, DVS 2205 published by the German Welding Institute. Only partial translations were published at the time the tank was designed, but the design philosophy was perfectly clear. The design procedure is described in Paper 5, and makes allowance – using safety factors, or derating factors – for holding dangerous chemicals, pinholes in welds and so on. The manufacturer of the tank, had hired an engineer to check and approve the design. The engineer performed misguided calculations, leaving the insurers to shoulder the considerable expenses of the clear-up.
So, having explained the failure, what were the consequences of the investigation? As is usual, several investigators had been instructed by loss adjusters acting for the two injured parties. The investigation described here was carried out on behalf of the tank manufacturer, and because liability was accepted, the role of the other investigators was limited. They did however, agree with the general thrust of the analysis.
A more serious problem – which arises whenever design defects are discovered – was the state of other tanks made to the same design. How many had been built and installed? What fluids were they holding? For how long had they been installed?
A range of tanks built the same way had been installed, but as it turned out, only relatively recently. The original investigator was asked by the manufacturer to inspect these installations, where tanks were holding fruit juices, soap solution and ferric chloride (FeCl3), an acidic fluid used for water treatment.
Using the given data, estimate the stress on the centre of the lower panel of some storage tanks when completely full with fluid of a given specific gravity. The tanks below are built like the tank introduced in Section 6.2, and are made from 12 mm thick polypropylene sheet.
(a) A 40 m3 tank holds detergent solution of specific gravity 1.2, has a height of fluid above the critical panel of 1.265 m, and has a radius of 1.91 m.
(b) A 25 m3 tank holds dilute detergent of specific gravity 1.16, has a height to the critical panel of 2.025 m, and has a radius of 1.6 m.
(c) A 25 m3 tank holds ferric chloride solution of specific gravity 1.5, has a height of fluid 1.8 m, and has a radius of 1.65 m.
Which tank is most likely to fail? What particular features would the investigator be looking for during inspection of that tank? What special methods might they use to help in this task? Assume specific gravity is equal to the density in units of gm cm−3.
The hydrostatic pressures in the different tanks will be given by Equation (9), where the densities of the contents can be inserted to estimate the pressure. The hoop stress σH at the centre panel can then be estimated using Equation (7), which relates hoop stress directly to the internal pressure exerted by the contents. The data can be used now to give answers for (a),(b),and (c).
(b) Dilute detergent
(c) Ferric chloride solution
So the last tank is the most seriously stressed in the centre of the lower panel, and should clearly be examined first. It may also be observed that the hoop stress is actually above that estimated for the tank containing caustic soda in the case study, so the matter is urgent. All welds need examining, preferably using white dusting powder to show any cracks in the key welds.
Although the tanks were relatively new, few were found with microcracks, essentially because none of the tanks had been filled to capacity. The cracks in the welds that were found were far from critical, being only millimetres in size. The national economy at the time was depressed, and the companies concerned had been on short-time work. Some of the other tanks inspected were much smaller than that at Warrington, so had not been stressed to the same extent. The most serious potential problem was found at a steel-wire works in the Midlands, where there were two adjacent tanks (Figure 78). One held caustic soda, the other ferric chloride. If the two tanks had split, the two chemicals would have reacted together with the evolution of heat, possibly causing a fire. The tanks were in a confined space, close to manned equipment, so the consequences of failure would have been serious. The bund walls had only been designed to accept the contents from a floor leakage, so would have been no help in a crisis like the Warrington failure.
No microcracks were found, however, because the tanks had never been more than half-full since installation. The factory was set in a narrow gorge and the only access to the tank building was over a small bridge incapable of carrying a full tanker. Needless to say, this and other tanks had to be replaced by tanks of the correct design.
Storage tanks continue to fail however, despite the publicity given by Papers 4 and 5. One failure tha occurred in 1998 involved a paint tank. It had been designed with a sloping floor, as shown in Figure 79. This internal cone was essentially to allow all the contents to be mixed within the tank and then drained away for further processing. It failed during the first time it was used, collapsing in a misshapen heap on the factory floor, letting the contents spill on the floor.
The weakness of the design is that the mass of the contents is taken by a single weld (arrowed in Figure 79), with no support at all for the inner cone. The weld had peeled away, failing progressively, so that almost all the weld had failed. The design was again at fault, with the absence of support recommended by the DVS 2205 standard, such as a steel frame resting on the ground, and bearing the weight of the cone and contents. Fortunately, the design was a one-off, being made by a small company for a specific contract, and had not been repeated elsewhere.
The two containment case studies were examples of problems capable of proceeding to litigation, but were resolved amicably by the insurers and their experts, who largely agreed about the causes. It is when the experts cannot agree, or come to quite different conclusions about causation, that litigation ensues. The case studies in this section will examine several problems that proceeded to litigation – and in one instance to trial – before settlement could be reached.
It is in the area of pipe and tubing that polymeric materials have become widely used in the last few years, often because they are more resistant to fluids, perhaps stronger than conventional materials, and certainly light and easy to install. Existing piping in materials such as steel, cast iron and ceramics is therefore increasingly being replaced by polyethylene, polypropylene and PVC.
Cast iron in particular was used widely in the Victorian period for water pipes, but it is susceptible to brittle cracking, causing leaking and loss of water supply. While water loss in the ground may be inconvenient, loss of gas is more serious, leaking cast iron mains having caused numerous explosions, fires and subsequent loss of life in the recent past. By contrast, polyethylene pipes are more flexible, so can accommodate imposed loads in the ground, and are much tougher or crack-resistant than cast iron. They can be easily welded and are easier to handle during fitment.
Polymers have also replaced many cast iron water supply systems for similar reasons, but problems can arise if the materials are susceptible to any chemicals in the supply. The particular problem we will examine in this section, arose from chlorine added for purification purposes. It caused stress corrosion cracking (SCC) of a plumbing fitting, which when it failed, led to substantial flood damage. The case led to litigation, which was only resolved by discovery of parallel problems in the USA.
Another critical application for tubing is in vehicle fuel lines, because if a break occurs, the proximity of a hot engine and electrical equipment can easily lead to a serious fire. The problem arose in the late 1970s after a serious incident in which a new car suffered fire damage, and other fires happened in the same model. The insurers commissioned a report that identified the causes of the problem, and a recall was initiated to replace the faulty fuel lines. Unfortunately, the recall was not fully implemented everywhere, and a serious accident in 1988 in the Republic of Ireland resulted in severe injury to two young children. The resulting High Court action revealed the nature of the problem, but discovery of key documents was resisted by the manufacturer, Fiat S.p.A. Subsequent investigation subsequently showed many fires in this particular model had occurred, and they were reconsidered in the light of the defective fuel lines.
One area of increasing litigation is that involving medical treatment that fails because of a product failure, and is an important aspect of medical negligence. As new materials are introduced into medical practice, the problems of poor material or design are highlighted because of the often severe personal consequences of failure. The human body is an aggressive environment for most materials, so care is needed in materials selection. On the other hand, such materials may fail for other reasons, and this was so in the final case study. The case also led to court action, but was settled just before trial when critical evidence regarding the mechanism of failure became available.
This case study involving litigation occurred when a small plastic fitting failed suddenly one weekend during November 1988, under a sink in the physics department at Loughborough University (Figure 80). The subsequent flood of water from the cold water main into the computer department immediately below is described briefly in Paper 6. Paper 6 is attached as a pdf which should be printed out (if possible) to gain the maximum benefit from the discussion of this case study (at least it should be kept open on the desktop throughout Section 7).
Click on the 'View document' link below to read Paper 6
The plumbing arrangements were the focus of attention by the loss adjusters and their expert (Figure 1, Paper 6). The acetal fitting was situated below the hot water tap, and was supported in a steel bracket, so that when the hot tap was turned on, cold water flowed through the large water heater to the tap itself.
It was noted that the heavy water heater was attached to the wall by only a single screw, which had come adrift from the wall, and placed the surrounding copper pipes under bending loads. The break in the fitting was just below the lower bend in the steel bracket, where the rising main supplied incoming cold water to the system.
Draw an FTA diagram to indicate all possible causes of failure that you think could explain the failure of the acetal fitting. The initial theory of the failure suggested the sagging water heater had overloaded the fitting, the load being transferred via the cold water pipe. Using Figure 1, Paper 4, sketch the load path – the path along which load is transmitted – between the water heater and the fractured fitting. From your diagram, comment on the credibility of the theory. Another suggestion could involve downward movement of the rising mains. How likely is this alternative hypothesis?
Figure 81 suggests some possible failure modes, including overloading by excess pressure from the water heater, overtightening of the joint and loading from the rising water mains pipe.
Figure 82 shows the load path between the water heater and the critical acetal fitting.
With the heater no longer fixed to the wall, the load is supported by the two copper pipes, and is shared roughly equally. The upper horizontal pipes would probably bend.
The lower pipe will transmit some part of its load to the upper junction of the plastic fitting, and it in turn will load the steel bracket. The angle bracket holding the fitting to the tap junction is made from steel and clearly capable of supporting the part of the total load transmitted to it. If breakage in the plastic fitting were likely, then it would be the upper rather than lower junction that should fail. As it was the lower part that failed, the theory is not credible.
Loading from a downward movement of the water mains pipe is more likely, because that could cause the key junction to be put in tension. However, the load is shared between two pipes, and the smaller pipe extension to the acetal fitting is likely to be less strained than the longer extension to the cold-water tap.
There must have been another reason for the failure. The next suggestion was that the junction had been overtightened by the plumber. The system had been fitted during renovation work some four years earlier. This is an interesting suggestion, because it seems to imply the fitting could be overtightened, and might have been a faulty design in allowing overtightening. It is of course a common way of producing extra stress on a joint, but what did the evidence show?
If screw joints are overtightened, there might be evidence from the rubber washers used in the system. They take any excess stress from tightening, so if the joint had been overtightened, then some damage might be expected. In fact, there was no evidence the washers had broken or distorted excessively at all (Figure 83).
But this is where the problem lay when it came to assessing who would pay for the substantial damages to the computers on the floor below. The loss adjuster's expert effectively pointed at overtightening as the cause of failure, possibly exacerbated by the ‘faulty’ screw attachment to the wall. So the loss could be recovered from the plumber or company who installed the system, and possibly the architect for planning the system in the first place.
The architects, or rather their insurers, refused to accept the blame, so proceedings were started by the university against the plumbing company and architects. At this point, the defendants naturally needed their own expert to examine the failure and produce a report – preferably someone with expertise in the failure of plastics materials. The failed acetal fitting was critical, but had not been examined in detail by the first expert. Why not? The reason or reasons are not known. It seems so obvious that the evidence at the heart of a failure, the cracked fitting in this instance, should be subjected to detailed analysis. However, some experts fail to make the leap into detailed inspection, perhaps because they do not have the expertise or laboratory equipment, and so on.
We have already seen the fracture surface in Figures 23, 24(a) and 24(b), where it was used as an example of how to start assessing the evidence of fracture during macroscopic inspection. The surface showed several old fracture origins, now concealed by debris deposited from the water supply, and implied from flaps left by crack growth (Figures 2 and 3, Paper 6). The detail shown in the latter figure suggests there were at least five different growth regions, with crack growth directions as indicated by the bold arrows. The white areas represent the fresh fracture induced by the loss adjuster's expert when the sample rolled off his desk.
When the areas adjacent to the fracture surface were examined in detail, one significant feature emerged: there were numerous subcritical cracks that appeared to have grown from weld lines in the threads and on the inner surface of the fitting (see Figures 84, 85), (see Figure 3, Paper 6). The weld lines appeared to be associated with flow lines in the outer surface of the moulding, suggesting moulding may have been faulty. Flow and weld lines are indicative of cool tools, so that flow of molten polymer in the mould tool is inhibited. That the fitting had been screwed up to produce a closure stress on the threads was indisputable as Figure 86 shows. So, what effect would such weld lines have when the thread was screwed tight by the plumber?
Evaluate the stress concentration at the threads of the fitting using standard diagrams already presented in this unit. The wall is 1.6 mm thick to the base of a thread root, and the root possesses a radius of curvature of 0.2 mm. What is likely to be the effect of having a weld line at a thread root? Assume a weld line is 0.1 mm deep. What light does the analysis shed on the way crack growth occurred? From your analysis, indicate whether cracks grew from the inside out or from the outside inwards.
The stress concentration at the thread root can be evaluated using Figure 40, Box 14, in a similar way to the calculation of SAQ 10(c). Using the values of wall thickness and radius of curvature of the root, then
r/d = 0.2/1.6 = 0.125
Reading from Figure 40 gives a Kt value of about 2.6, a relatively low value. However, any weld line within the root will magnify the stress further. Suppose, for example, the stress concentration is about 3, the combined effect is to give a stress concentration of 7.8, the product of the two independent stress raisers.
The net effect of the several stress raisers will operate on the final closure stress of the joint made by the plumber, but the evidence from the washers showed there was no excessive force applied. There would have been only a small closure stress on the threads and hence only a relatively small magnified stress at the tips of the weld lines at the thread root. It is interesting to note that subcritical cracks within weld lines were also found on the inside surface of the failed fitting near the fracture. Although the stress concentration will be lower – there is no thread root – it is possible the original cracks grew from inside the fitting as well as from the outside.
However, the overall stresses would have been small and there must be another explanation for the early failure of the joint. Whether or not the cracks grew internally or externally thus remains unknown. The fracture surface itself is of little help.
So tiny weld lines could concentrate the tightening load if any were at the root of a thread. One common observation of failures from screw threads is that failure is often at the last thread but one – a feature confirmed by theoretical analysis of the stress-concentrating effects of thread forms. Thread failures are themselves common because threaded joints and connections are one of the most common ways of joining components.
However, there was no evidence of an excessive closure stress made by the plumber when finally tightening the joint, and the cause of the failure remained unknown at this stage of the investigation. What was clear was that there was no evidence supporting the proposal that the joint had been overtightened – therefore the plumber was not to blame.
ESC is a recurring problem where tough polymers are simultaneously stressed and exposed to organic fluids that can initiate brittle cracks. While many polymers, such as polyethylene and polypropylene, are insoluble in most common organic solvents, some liquids are so aggressive they can be absorbed and swell the surface layers. Swollen polymer is less resistant to crack formation, so that microcracks can develop on exposure. Further exposure can lead to crack growth, which, when it reaches a critical length, leads to catastrophic fracture of the product.
ABS is a tough plastic susceptible to such cracking, often under unusual circumstances. However, there are many organic fluids that can cause ESC in a surface under strain. This is what appears to have happened when an ABS compressed air line suddenly exploded in a factory in 1998. Fortunately, nobody was injured, although there was substantial damage to property. Examination of the large fracture surface showed a characteristic pattern of cracking on the inner wall (Figure 87).
The edge of the pipe showed a crenellated pattern, where numerous subcritical cracks had formed at the stressed surface and grown into one another. The largest crack had grown catastrophically, causing the explosion. But what had initiated the cracks?
When examined closely, the inner surface of the pipe showed faint traces of a fluid contaminant that was always associated with a brittle crack (Figure 88). One possible source of the fluid could have been the oil used in the compressor motor being blown suddenly, for some reason, into the otherwise empty pipe just prior to the explosion.
Like SCC, the failure mode is not unusual with stressed polymers, perhaps because not enough is known about the phenomenon together with the increasing variety of fluids used in industry. Most manufacturers supply lists of compounds that can cause cracking, but the list is inevitably out of date as potentially harmful reagents are coming into contact with strained surfaces.
The effects are determined by the concentration of fluid in contact with the polymer surface and the degree of strain. There is usually a critical value needed to initiate cracking. A standard test used for testing potentially harmful reagents is the Bell telephone test where bent strips of polymer are exposed to the reagent under standard conditions.
What can be done about the problem? One solution is to coat the strained surface with a resistant polymer, another being to use a grade of higher molecular mass. The latter is always of greater strength, but will increase process costs owing to its greater viscosity.
One additional factor is the internal pressure from the water supply. Although it had not been measured directly, the specification of the fitting allowed for a pressure of up to 25 m head of water. The pressure p would have imposed a hoop stress σH on the wall of the fitting, but what magnitude would it have been? The pressure can be calculated using Equation (9), and the hoop stress with Equation (7), Box 19.
The hoop stress, the largest stress acting on the pipe wall from the water pressure, in the system is given by
This stress is small compared with the nominal tensile strength of acetal material, of about 70 MN m−2. Although moulding could reduce the strength somewhat – as seen in the case of the failed radiator – it was difficult to see how the combined effects of internal pressure and screw thread stress raiser could initiate crack growth, and ultimately, catastrophic failure.
What could have created the weld lines in stressed areas of the moulding? A detailed survey was made of new mouldings supplied by the manufacturer. They were smeared with graphite powder to highlight the flow and weld lines. There were considerable variations between the twelve mouldings examined (Figure 89). It was concluded that there was a pattern to the flow lines. The set had probably been made in two batches, each producing a different flow line pattern. Weld lines did exist, but were much less serious than in the failed sample. Moreover, the failed sample showed weld lines where none could be found on any of the new mouldings, especially in the critical threaded areas of the lower inlet pipe. The failed sample had been made in a different batch, and may even have been a maverick sample.
One failure mechanism that has been known for many years is known as environmental stress cracking (ESC). Certain fluids can cause brittle cracking of plastic products, even at low imposed stresses. One of the first examples to be discovered occurred in polyethylene when exposed to strong detergent, such as an ionic soap known as Igepal. It was also the probable cause of an explosion in a compressed air line discussed in more detail in Box 20.
Another related failure mechanism is stress corrosion cracking (SCC), where the fluid interacts chemically with the polymer surface it contacts. One example was referred to earlier in Box 4, where radiator seals failed by attack from hot water in the central heating system. There were several subcritical cracks in the failed washers, a characteristic feature of both ESC and SCC failures. For attack to start, something must open the crack. It takes two main forms.
Small stresses or strains imposed on the system.
It is also known that the more aggressive the attacking agent, the lower the critical strain needed to grow cracks.
In its original form, when polymerized experimentally, polyoxymethylene was not a useful plastic. It easily degraded back to monomer, in this case formaldehyde. This could occur when simply heated, in an injection moulding machine, for instance. Two strategies were adopted to improve the stability of the material to make it suitable for products: copolymerization with another monomer, and end-capping every chain. Both versions of the polymer are available commercially, the failed fitting being made from the latter.
However, both the commercial polymers are susceptible to attack, especially by oxidizing agents. It was known that acetal could not be used in swimming pool plumbing, where chlorine levels can be high. Chlorine is a powerful oxidizing agent, which is why it is used for water purification – it attacks bacteria. Could the much lower levels present in potable water have attacked the material?
While the various expert reports were being digested by the several parties now in the action – university, plumbers, architects – one of the investigators happened to come across a report in a technical newspaper that suggested there could indeed be a problem with plastic plumbing materials. The article is shown in Figure 90, and is dated 18 March 1991. It reports the settlement before trial of a case involving domestic hot water pipes, where polybutylene pipes had been connected with acetal copolymer fittings. They had failed and caused flooding. The newspaper reported a class action, an action taken by the affected families against the manufacturers and suppliers of the materials. Both the pipe and fittings had suffered extensive cracking on their inner surfaces, a problem expert opinion considered to be caused by chlorine and dissolved oxygen in the water.
A further class action in Texas came to trial in 1992, and resulted in a large settlement for the numerous plaintiffs. This case (Babb v. US Brass, Shell, Hoechst-Celanese, DuPont) also attracted the attention of the expert acting on behalf of the plumbers in the UK case. Contact by the instructing solicitor with the American attorneys produced not only copies of the expert reports in the case, but also the transcripts of the trial – where the expert evidence had been tested in public. It was clear from those expert reports that even low levels of chlorine down to 0.3 parts per million in the water could cause serious stress corrosion cracking of acetal.
Such chlorine levels can occur in drinking water, so it seemed as though the problem of the cracked acetal fitting may have been caused directly by chlorine. A check with the local water company produced extensive analyses, both from 1993, and from when the fitting had failed in 1988. The records showed free chlorine was variable at different sampling points, but could rise as high as 0.9 parts per million. One reason for high levels was given as being caused by slugs of chlorine being added by the water company when there were accidental breaks in the pipes – a precautionary measure to prevent contamination.
A direct experiment was then carried out on the failed fitting using the EDAX facility of the scanning microscope. Analysis of the surface showed the presence of chlorine in significant quantities. Presentation of the new evidence to the other parties in the dispute produced a rapid settlement of the action, with the parties walking away from the action, and bearing their own costs.
The USA actions were appealed by the defendants, but were lost in the Texas Court of Appeal, and the many plaintiffs received full compensation. This was incidentally one of the largest class actions – outside automobile actions – in the US and ultimately cost the companies millions of dollars in damages awarded plus litigation costs.
What did the failure at Loughborough University show? There appeared to be no widespread failures of similar fittings in the UK, so it would be reasonable to conclude that the particular fitting was indeed a maverick. Perhaps it was supplied by mistake, for the same reason that the faulty radiator reservoir was fitted to a new car in the previous case study.
The failure could be attributed to low levels of chlorine in the water supply, which led to rapid creep rupture at weld lines on the inner bore adjacent to the threads. The several cracks thus initiated eventually merged as they grew slowly. Crack growth probably slowed down as the closure stress was relieved by their formation. During this period, water leakage was slow and allowed deposits to grow on the fracture surfaces, helping to stem the leak. This helps to explain why the leak was not noticed, or if noticed, dismissed as insubstantial.
Some trigger over the weekend of 6–7 November 1988 allowed the crack to split open and the water damage resulted. Movement of the mains inlet pipe, or perhaps water hammer – a sudden surge in water pressure often caused by valve a closing or opening – could have provided the trigger (Figure 5, Paper 6).
The case shows how important it is to be aware of problems that may have occurred in other countries using the same material in a similar way. The USA case also revealed, in the expert reports, how the manufacturing companies had actually tested the materials with low levels of chlorine long before they were promoted as fit for water plumbing fittings. The laboratories in those companies had shown that levels as low as 0.3 ppm of chlorine could cause cracking and hence destroy the integrity of the materials. Why that information was not acted upon before they launched the plumbing fittings and pipe remains unknown.
Polybutene has been withdrawn from the USA market by the makers as a direct result of the US litigation, although acetal remains, and is widely used for fittings both in the USA and in the UK. Now, though, manufacturers should be fully aware of the possible problems that can ensue if low quality mouldings are supplied to users.
The next case study involves extruded tubing used for fuel lines under the bonnet of a saloon car in the late 1970s. The car, the Fiat Mirafiori, was brand new at the time, but several were damaged from fires that were not the result of an accident. This raised issues of consumer safety, quite apart from the warranty damage involved.
One fire occurred when a new car owned by the late Sir John Gielgud, an eminent English actor, was travelling along the Embankment in London. Smoke came from the engine compartment, and although the fire was extinguished, substantial damage was caused (Figure 91). Examination of other fire-damaged cars by alert insurance inspectors showed the fires were probably caused by a leaking fuel pipe near the engine.
The matter was apparently solved with a recall of the cars and the replacement of the pipes. But a fire-damaged Mirafiori re-appeared in the early 1990s in a case before the High Court in Dublin. It raised the problem of the dissemination of failure studies, especially in an international market where identical or very similar products are supplied to different national markets. It also focused attention on the importance of quality control during product design and development, especially where consumer safety is paramount.
Examination of the material of one of the pipes from the fire-damaged cars showed it to be nitrile butadiene rubber (NBR), which is petrol resistant, but an intact sample of pipe showed ozone cracks (Figure 92). Ozone gas is extremely reactive and is produced in the atmosphere by the action of sunlight on car exhaust fumes – the gas levels rise in the spring and summer. However, the gas is also produced near electrical equipment, especially during sparking, so concentrations can be high under the bonnet of petrol engines.
Ozone cracking is easily started at bends in susceptible tubing (Box 21). Crack growth occurs laterally, and penetrates from the crack tip into the interior of tubing.
One of the most serious ways many rubbers can be attacked by their environment is from ozone gas, normally present to a small extent in the atmosphere in 1–10 parts per hundred million (pphm). It is a reactive form of oxygen gas. Oxygen gas is diatomic with the formula O2 In ozone, an extra oxygen atom becomes attached by an input of energy to form a triatomic molecule, O3.
When air is bombarded with energetic electrons or sunlight, ozone can be formed. In the upper atmosphere, it forms a protective shield for life on earth by absorbing harmful short-wave ultraviolet radiation, but at ground level it is a polluting irritant. The most serious pollution incidents occur on bright, sunny days in the presence of car fumes.
Ozone is a powerful oxidizing agent, and is used for water purification, for example. However, it will attack and degrade many rubbers by combining with the double bonds present, and splitting the polymer chains. The net effect is that the polymer chains become smaller and the material loses its strength. The way it loses strength is by way of cracking from the surface at right angles to the applied strain (Figure 93).
Only small strains of about 1 per cent are needed to initiate cracks. Strains can occur simply by bending pipe or tubes slightly. Attack is concentrated at the crack tip where fresh polymer is exposed to the gas. The rubbers most at risk of attack are those with double-bonded chains, such as natural rubber, poly butadiene, and any copolymers, such as NBR.
Ozone cracking was commonly seen in old tyres, being easily distinguished from oxygen cracking by the deep and highly oriented pattern of cracks. However, it is rarely seen today, not only because tyres are discarded earlier in their lives, but also because anti-ozonants are added to rubber compounds to give protection against the gas.
With the Mirafiori engine design, the material cracked, eventually punctured, and leaked petrol on to the manifold. When the petrol hit the hot engine manifold, it vaporised, and a small spark from an electrical circuit could initiate a fire.
In addition, a problem with another part of the fuel system revealed a thermoplastic material had been used for the return pipe that passed over the engine, returning unused fuel to the petrol tank. Owing to its poor abrasion resistance, this pipe could also leak if any part contacted the engine manifold, where rubbing caused a hole to develop (Figure 94). The pipe was an extrusion of nylon 11, and possessed a melting point of about 176° C. The fire in Gielgud's car had probably been caused by abrasion through the wall of the return pipe rather than ozone cracking of the feed pipe.
As several cars had been damaged by such fires, it seemed clear that both unprotected NBR fuel pipes and nylon return pipes had been fitted widely, and so a recall of the model was carried out (Figure 95).
The recall was carried out in the early 1980s, so it came as a surprise when the original investigator was approached by lawyers involved in a High Court claim for damages in the early 1990s. A serious fire in the Irish Republic in the late 1980s in a rather old Fiat Mirafiori had caused serious injuries to two young children when in the rear of their mother's car. She had been shopping, and when arriving back home, had removed the ignition keys to open her front door. As she turned back to remove the children from their seat belts, the car interior burst into flames. Fortunately, the children were rescued by neighbours, but not before they had suffered serious burns. The resulting court action claimed substantial damages for their continuing care, and the pain and suffering.
The accident was unusual in that the fire had occurred in the passenger compartment of the car rather than the engine compartment – where the original problem had been discovered. What was the likely cause? Unfortunately, most of the evidence had been lost in the fire, the car being a burnt-out shell when examined later.
There were two possible causes: a faulty heater near the dashboard, and the plastic return pipe to the petrol tank at the rear of the vehicle. The seats were filled with inflammable polymer foam, but the rapidity of the fire suggested a petrol leak and ignition from a spark. Ignition could have been from the faulty heater, although it is well known that static sparks can easily ignite gas or fumes (Box 22). The fuel return pipe passed through the interior of the car, and could have leaked petrol just before the fire. It would then only need a small spark, perhaps when the woman left the car or closed the door, to ignite the fumes. A small fire could then become an inferno as the liquid petrol caught fire, and the fire grew into the seats. Such a scenario seemed feasible, but lacked corroboration in detail, mainly because most of the evidence had been destroyed in the fire.
Although a common phenomenon, static electricity can be a safety hazards because of the energy released during discharge. It is caused by contact between dissimilar materials, where friction and wear break chemical bonds and so release electrons. The materials must be non-conductors, so allowing build-up of charge on the surfaces. An old example was said to be rubbing an ebonite rod on a cat's fur, but a more familiar example would be rubbing an inflated rubber balloon on fabric. The balloon will then stick to a vertical wall as if by magic.
Static electricity is of course the source of lightning during thunderstorms, and although the mechanism is not known in detail, is thought to be created by friction of raindrops against one another – if true, it breaks the dissimilar material rule.
Static electricity is used commercially in many different processes, for instance paint spraying where the paint particles are charged and are attracted to the surfaces to be painted.
Certain semi-conductors such as selenium will hold a static charge, except when exposed to light. So if an image is shone onto a statically charged plate of selenium, a replica of the image is formed on the plate. If a polymer toner dust is added to the plate, it is attracted to the image and can be developed by simply heating the particles.
The forensic method known as ESDA works on a similar principle, but detects the faint impressions left on paper when the sheets above are written upon. The method has been useful in revealing changes in statements or notebooks and has been crucial in exposing numerous unsafe convictions in the criminal courts.
The build-up of static can be a serious problem in atmospheres that contain inflammable gases. There were a series of disastrous explosions in the early 1950s in operating theatres, when ether-laden atmospheres were ignited by static sparks suddenly leaking to earth. But why should this have occurred when ether has been used for years in hospitals? The build-up of static was probably caused by the introduction of many new synthetic fibre textiles into clothing and hospital apparel under dry conditions. Nylon fabric, in particular, is known to create static. When a nurse or surgeon came close to a metal earth, such as a window frame, the static build-up created a spark that ignited the atmosphere. The solution to the problem was to provide conducting floor materials to allow any charge build-up to leak away harmlessly to earth. The addition of cotton to some synthetics also created safer fabrics.
There can also be problems in dust-laden atmospheres, especially if the dust is fine enough and flammable, as is the case with flour and resins. Conveyor belts are ideal static electricity generators, especially if the rubber belt rubs against another material. The sudden discharge creates a spark, which then ignites the atmosphere. Two remedies can help prevent the problem: increasing the humidity of the atmosphere allows slow leakage of static – conversely, dry atmospheres help cause build-up – and using conductive materials. Rubber can be filled heavily with carbon black to provide a conduction network that provides a leakage path for the electrons.
By the time the investigator's original report was requested, the case had reached an impasse. In personal injury actions, the small claimant often faces the problem of discovering details of manufacturing processes, quality control procedures, and previous incidents, from the defendant companies. The problem of discovery from Fiat S.p.A. in the Republic of Ireland had led to a so-called striking-out claim against the company, because they either could not or would not release the required information to the claimants’ solicitors. The latter pursued the information independently, and came across the public information of the recall in the early 1980s. The trail led to the insurance company, which had unfortunately disposed of the original independent report on which the recall was based. However, they did have a record that the report had been commissioned, so they were able to make contact with the original investigator. Fortunately, he had retained all samples and reports from the first accidents, and was able to supply them direct to the legal team in Ireland. The report was supplied as an affidavit to the court, and the defendant witnesses – managers for litigation and engineering from Fiat, Italy – were cross-examined on the evidence of the fuel line problem.
The claimants succeeded in their action for striking out (Figure 96 shows a report made during the case), but the case went to appeal on the legal principle of striking-out the defence. Striking-out effectively meant Fiat could not succeed when and if the main claims went to full trial. Although Fiat won the appeal, the claim was settled before trial with a substantial award to the injured children.
Owing to the extensive publicity in the Republic of Ireland, the case revealed a series of fires in this model, and further claims were lodged after the successful settlement of the Murphy children. In hindsight, it seems remarkable that greater care was not taken during testing of the new model.
The radiator case in Section 5 showed how new cars are now driven in realistic conditions before launch to reveal any latent defects to the design team. This is why FMEA methods are so important during development, and why it is important to know what previous problems have occurred in such simple and low cost items as fuel pipes. The same investigator followed up the original study by examining fires in new Ford Cortinas, for example.
Car fires continue to plague many new models, perhaps because cars are so vulnerable within the engine compartment. However, reinforced fuel pipe is readily available to designers, made from resistant and strong materials, and there is no reason why such pipe should not be specified during the design phase. Ozone-resistant elastomeric pipe is also widely available, so this particular problem is not insuperable. Many synthetic rubbers, which do not possess vulnerable double bonds, have been made available commercially within the past couple of decades, for example.
Car fires are particularly distressing in their effects, and the wider problem was recently (1999) highlighted in the USA, when the largest ever damages were awarded by a jury in Los Angeles. The car fire was caused by a rear collision that damaged the tank, and was not dissimilar to an earlier cause célèbre, fires in the Ford Pinto.
The case study of the Fiat Mirafiori illustrates the importance of cooperation between regulatory bodies in different countries, so that they are all aware of design problems with specific products, and can take appropriate action. It also illustrates why it is so important the basic engineering message of choosing components fit for their function is taken seriously, so that problems do not recur, and the consumer does not suffer as a consequence of ignorance or incompetence by the design team and the regulatory bodies.
An incidental issue is one of preservation of records and samples, so that the lessons of particular investigations are not lost. Indeed, publication achieves this purpose, because then the information is recorded permanently for public consultation.
It is true to say medical technology has developed at a breathtaking pace with the introduction of new NDT examination methods for exploration of the inner workings of the human body, such as NMR, X-ray tomography and advanced ultrasonic scanning. While vital for diagnosis of disease, it has also stimulated the development of new surgical methods of intervention.
Doctors treating heart disease have many options for treatment. Angioplasty, for example, involves threading a catheter upon which a plastic balloon has been wound, into an artery. When the catheter has reached a congested part of the artery, where fatty deposits are restricting the blood supply, the balloon is inflated and the deposits flattened. Such a procedure has been further developed by enclosing the balloon in a small metal wire cage, called a stent, so that when the balloon is inflated, the stent deforms plastically and remains in place to support the artery permanently. A key part of the procedure is being able to observe the exact point of placement using X-rays or ultrasound.
Developments in medical technology have not been without problems however. Product failures have been investigated, and some of the failures have been widely publicised within the medical profession, which has led to improved designs. Other failures, such as with breast implants and hip joint replacements, have received widespread public attention, either because of litigation or because action has been taken by official bodies (such as the Medical Devices Agency, the government body that monitors failures of such products). Many new materials, as well as conventional materials, have been introduced, with mixed results, as the next case study shows.
Catheters are small bore tubes by which fluids such as serum, blood, saline solutions, or drugs are supplied to a patient before, during, or after surgery or other medical treatment. A variety of polymers have been used in the past, such as plasticised PVC, but such materials can present the danger of plasticiser leaching into the human body.
Thermoplastic elastomers (TPE) offer a range of properties without the risk of leaching, so they have been adopted by many suppliers to hospitals. The case concerns a new type of TPE introduced in the 1980s, where the elastomeric properties were provided by short-chain poly-oxides, and the stability provided by physical cross-links of nylon segments scattered along the main chains. They co-crystallised into so-called hard blocks and so provided anchor points of stability for the material.
The TPE offers benefits for catheters, because it can be processed easily using conventional plastics processing methods, yet behaves like a vulcanised rubber. The stiffness can be modified simply by varying the hard-block content, producing at one end of the scale, a soft material, and at the other end a hard material, which are both tough and stable in use. Such a range of properties is not available either in conventional plastics or vulcanised rubbers.
The product in this study is a catheter, about 1 metre long and with an outside diameter of 1 mm. One end of it had been sealed and provided with three small holes for the infusion of drugs into the patient – that is called the distal end – as shown in Figure 97. The other end – called the proximal end – was fitted with a socket into which a syringe could be inserted by the anesthetist to provide the appropriate drug.
The catheter-plus-syringe is used in childbirth, where the mother requests an epidural anaesthetic, which effectively eliminates any pain from the lower half of her body. The injection into the spinal column is carried out by an anesthetist by inserting a heavy duty hollow needle – a Tuohy needle – through which the distal end of the catheter is fed. The drug can then be drip-fed through the catheter into her spinal fluid. The whole device is provided in a sterile, hermetically-sealed pack ready for use by staff when needed.
In one such operation, on 18 August 1990, all seemed to go well, with the delivery of a healthy boy to a Mrs K. The catheter was found to have broken at the proximal end, but was mended satisfactorily, according to a witness statement from an attending nurse. Removal of the catheter at the end of childbirth occurs by withdrawal of the needle, and gently pulling the catheter until it slides out neatly. It should not be withdrawn through the needle, because the sharp edge of the hollow tip could cut the catheter and leave a chunk in the patient.
But something went wrong with the removal from Mrs K: the distal end sheared off at the proximal side hole and a small fragment was left in her spinal fluid. After consultation, it was decided there was more risk in attempting to remove the fragment than leaving the piece where it was. The fragment was small, sterile and apparently presented no further risk to Mrs K.
The patient was disturbed by the decision to leave the fragment in her spine, and decided to bring proceedings against the hospital. The first action of her solicitor was to engage an expert to examine the remains of the catheter, which had eventually been passed on to the manufacturer of the device.
It was clear the distal end was the key to determining the cause of failure, and the expert duly examined it in the offices of the solicitor representing the manufacturer. However, he was not allowed to remove the sample for microscopic examination, and had only a short period to examine it with a hand-held lens. In his opinion the failed distal end of the catheter showed traces of knife or score marks running across the failed surface. It had probably been withdrawn incorrectly by the anesthetist.
The matter rested there for some time before the hospital engaged its own expert. In the interim, an expert had also been hired by the manufacturer, who was also a party to the dispute (in the event that the catheter itself proved faulty). The three experts needed to re-examine the failed end to provide firmer evidence of the failure mode. The failed device was provided to the experts, and given the 1 mm diameter size of the failed end, they decided to use a scanning electron microscope to provide clear images.
Rather than coat the end, it was agreed to examine it using the then relatively new technique of environmental SEM (ESEM), so that this critical piece of evidence would be preserved intact (see Box 23). The method produced an image that was capable of detailed analysis (Figure 98). The interpretation of the failed end by the expert acting on behalf of the hospital is shown in Figure 99.
The main problem with conventional SEM is the need to coat the sample surface of non-conductors with carbon or gold so as to conduct away the incoming electrons. If this is not done, they build up on the sample, and inhibit image formation. A very high vacuum is needed in conventional SEM because air molecules scatter and absorb the electron beam. A recent innovation, however, allows a small bleed of gas to pass over the sample being examined without entering the main column of the instrument (Figure 100).
As the primary electron beam hits the surface, any electrons that stay on the surface are neutralised rapidly by reaction with positive ions formed by interaction of the primary beam with the gas molecules bled into the microscope. This enables the electrons to be carried away from the surface of the sample, so preventing the unwanted build-up.
There are several versions of the new method available. In the version used for the failed end of the catheter – so-called low-vacuum SEM, or LVSEM – the pressure is kept rather low at about 30 Pa. But higher pressures are available in some instruments, rising up to about 2000 Pa.
The prime interest in these instruments is for observing living things, or materials that would deteriorate rapidly at low pressures. Water absorbent fibres and wood lose water rapidly at low vacuum, so suffer damage. Care is still needed for all polymers however, because the highly energetic electron beam – accelerated through 20 000 volts typically – can itself cause direct damage to samples by chemical reaction with the polymer chains.
Examine the failure surface map (Figure 99) produced by one of the experts. Summarise what the map shows in terms of the behaviour of the normally tough TPE material. Are there any traces on the failure surface that indicate the catheter was cut by the edge of the Tuohy needle?
Indicate the stress concentrations you would expect from the distal end of such a catheter if it were strained in tension along its major axis. Attempt to provide a quantitative estimate of the net stress-raising effect. The expert has marked apossible origin on one exterior surface of the sample. Is this a credible interpretation, given the presence of several severe stress concentrations?
The failed distal end of the catheter shown in Figure 98 exhibits both brittle and ductile behaviour. The brittle nature of the areas either side of the hole in the side wall is shown by the large flat zones, which also show contamination from dust and other debris on these relatively featureless zones.
There is, however, clear evidence of ductility in the remaining parts of the failure surface. In Figure 98, the quadrant between 7 o'clock and 11 o'clock shows two tear zones leading to a ductile tip. Moreover, there are traces of fibres pulled from the surface at 3 o'clock, adjacent to the edge of the side hole. The overall surface appears unusual for a normally tough and ductile material.
When a tough polymer is cut by a knife in a single event, tiny imperfections on the blade usually produce lines running straight across the surface. Also, a ductile tip is often created where the blade edge leaves the surface. There appear to be no parallel markings on the fracture surface. Alternatively, it might be argued that a brand new needle would have no such imperfections, and that the absence of such marks could not be taken as evidence that the sample had not been cut. On the contrary, the large flat zones would be what one would expect from a single cut.
The major stress raisers exposed in the failed surface of the catheter are:
the large hole in the side wall of the catheter;
two inner corners to the hole.
The inner corners were probably created when the hole was made during manufacture, probably by a hot needle. The standard effect of a hole is to produce a Kt value of about 3, although in a sample like this, the effect may be larger because the size of the hole is comparable with the inner bore. The effect of the corners can be estimated using Figure 40, Box 14. The relevant dimensions can be estimated directly from the failure surface map or photograph. Taking r to be about 2 mm, and d to be about 24 mm gives
r/d = 0.08
So, reading from Figure 40 gives
So the net stress concentration is at least 3 × 3 = 9.
The problem is that there appears to be no clear origin either at or near the edge of the hole, which would be expected from this calculation. On the contrary, the only possible origin shown on the surface appears to be some distance away, on the outer edge of the tube, and adjacent to a large flat zone. The failure surface shows apparently contradictory features; it appears not to offer an easy explanation.
The failed surface was clearly rather complex, showing areas of both brittle and ductile behaviour in a nominally tough material. The large flat areas could be interpreted as cut areas, with a ductile tip representing the point at which the sharp edge of the needle left the tube. This would fit with a supposed entry point on the opposite side of the catheter, near the proximal side hole. On the other hand, there were no traces of parallel markings from the minute defects in a knife edge. Further investigation was clearly needed, but what was the next step? Comparison of the failure surface with other parts of the failed catheter, such as the failed proximal end, and a deliberately cut catheter could be useful.
In an effort to produce further evidence of the state of the tube, conventional SEM was used to examine the state of the catheter elsewhere. Several other areas of the intact catheter showed traces of brittleness (Figure 101). Taken from a broken edge well away from the failure surface, the conventional SEM micrograph shows a completely brittle fracture surface, with hackles and sharp edges characteristic of completely brittle behaviour – just as you would expect from, say, broken glass. In addition, a sample of tough new catheter was deliberately cut with a Tuohy needle, producing the failure surface shown in Figure 102. The cut surface shows some features that are common to the failed catheter end, such as two areas of ductility and a flat area. However, there are parallel marks showing defects from the edge of the needle.
If a product shows traces of brittle cracking when it is normally tough and ductile, it is essential to check the material has not degraded in any way. So samples of new and failed catheter were subjected to DSC, FTIR, and even NMR in an effort to provide some evidence of its quality. No degradation was found at first, although the exercise was useful in determining the composition of the block copolymer. As was found in the case of the cracked radiator seals described in Section 5, minute traces of degradation are often easy to miss using spectroscopic methods. Attack is concentrated at the tip of a crack or cracks, so the spectra are swamped by normal polymer.
However, there was one piece of evidence that was obtained using FTIR microscopy. This is a method where a narrow infrared beam is passed through areas of the sample selected while in an optical microscope, so that very small parts of a sample can be analysed. The result of one such experiment from the expert representing the manufacturers is shown in Figure 103, where the failed catheter is shown by the lower curve, and a good catheter in the upper curve. Quantitative analysis from the base lines shown indicated the presence of significant amounts of esters and other products in the material.
An inspection of the scientific literature on the worldwide web produced an important article by a group of French chemists, the abstract of which is shown in Figure 104. Their work showed photo-oxidation by ultraviolet rays attacks the amorphous or elastomeric parts of the molecules, producing esters – among other degradation products. As UV attack cuts the main chains, the strength can drop rapidly where exposure has occurred.
Tensile testing of the material also showed the strength of new catheters, but several entirely brittle failures were also obtained from the length of failed catheter still available. As before, the experiments were conducted with all the experts present, so that later, when the case came to trial, there could be no argument among the experts about the validity of the experiments – arguments are a possible problem where experts perform their own research in isolation. Figure 105 compares the load-elongation curves from a good and the failed catheter.
At this stage, the experts still disagreed about the cause of failure, but the possibility of degradation had been strengthened by the observation of traces of foreign ester groups in FTIR microscopy. Discovery from the makers of the catheter produced information about extrusion and sterilisation, enabling a traceability diagram to be constructed (Figure 106). The diagram shows at left the unknown dates of the various events in the history of the failed catheter, while the right-hand column shows the various unknown conditions of the various manufacturing processes used.
Exposure to sunlight is unknown at all stages, but could have been considerable. No quality control procedures had been revealed that could have caught any degraded material. Excessive heat during extrusion could also have damaged the material. Most significant of all in the sequence is the gamma radiation exposure step, the final step where the product was sterilized. The problem arises of further degradation here if earlier exposures to sunlight had occurred. Gamma radiation is powerful, and would have exacerbated any damage to the polymer chains. But no other failures of such catheters had occurred, so what was unique about this catheter?
Consider the traceability diagram of Figure 106, and determine any stages where photo-oxidation could have occurred. Supposing you were the expert acting for the manufacturers, what action would you take to check the degradation hypothesis? What quality records are likely to have been made at the time? What counterarguments would you use to challenge the degradation hypothesis?
The traceability diagram of Figure 106 shows several stages when photo-degradation could have occurred. Attack at the granule stage is unlikely, and the degradation would be distributed uniformly through the product by melting and mixing in the injection-moulding machine. Degradation is more likely on extruded tube, and should include any stage when the product could be exposed to sunlight. Degradation could have happened in any of four ways – when it was:
stored after extrusion;
cut into lengths and heat processed;
taken from the package early at the hospital;
taken from the package prior to the childbirth.
Acting on behalf of the manufacturers, the expert would check the internal quality records, which were never made available in discovery. The tubing was actually extruded in Ireland, so the search for records would start there. Quality records would include extrusion machine printouts that recorded temperatures and pressures for the specific batches in question (FLA 234/5). Although an extrusion problem is unlikely, it cannot be excluded. The product is made in batches, and there would be a period when non-specification tubing would be produced at the start-up of the machine – analogous to the problem in injection moulding that probably created the faulty radiator reservoir.
The most important quality records needed for examination would be the testing of small samples of tubing taken at random during production of a batch. The expert would be looking for any drop in mechanical strength – if such a test had been employed. There would also be, no doubt, an internal standard that specifies the frequency of testing and the kind of testing to be used, and at what stage the sample would be taken. There would probably be a random check after packaging, and possibly after the critical radiation sterilisation stage.
The expert would probably start by suggesting any exposure was caused after the supply of the catheter pack by the manufacturers, perhaps by a nurse opening the pack some time before it was needed. It might then have been left on a window ledge exposed to sunshine. Brittleness was confined to specific zones, and most of the catheter was tough – otherwise it could not have been used at all. It suggests accidental exposure by the hospital itself rather than a manufacturing fault.
But is pre-use degradation the answer? The theory is open to attack from several angles. Firstly, the catheter had probably suffered degradation in the period since the accident, because little attempt was made to protect the device from sunlight – although this point goes against the manufacturers, because they kept the device for several years following the accident. Secondly, could the catheter have been exposed to solvents, causing ESC or SCC? Hospitals use many organic solvents that might have attacked the material at isolated zones along its length, if solvent had been spilled. The expert would ask for further details from the attending nurses, and new witness statements to clarify the matter. No doubt, they would also ask the manufacturers for details of any known cracking or crazing agents, and then cross-check with the new witness statements. There is clearly some scope to attack the pre-use degradation theory.
One possibility that could have caused a rogue length of tubing could have been exposure of a coil of extruded tube to sunlight, shortly after manufacture in the Republic of Ireland (stage 2 of Figure 106). Exposure to sunlight could have damaged just a few outer layers of tube, just enough to initiate photo-oxidation and leave damaged chains. It is also important to appreciate that such a catheter would incorporate no additives to protect the material against such attack – something that is normal for other, non-medical products. The reason for using virgin material is to prevent leaching into the patient.
A final meeting of the three experts was fraught, because they could not agree over the balance of causes that led to final failure. The expert for the claimant still maintained that the anesthetist could have cut the catheter, although support for his case was weakened by detailed analysis of the fracture surfaces (which he recognised). The expert for the manufacturer was, however, resistant to any problems with the catheter as supplied to the hospital, and proposed a joint report be prepared for the court. This is an option courts do prefer, especially as it means the judge has less reading before the trial. The issues raised by expert evidence are frequently just as complex as shown here, and at the end of the day, the court has to rule on which opinion it prefers. With one agreed report, argument over causation is limited. This might be an easy task for just two experts, but becomes more difficult where three or more are involved. Two of the experts agreed to produce a joint report, while the third – acting on behalf of the hospital – produced an independent report. Alternatively, courts require an agreed statement of what is agreed and what is disagreed between the experts. It was not produced in this case, owing to a breakdown in relations between the experts.
There remained one final issue, which puzzled the expert acting on behalf of the hospital. That was the problem of the failure itself. Some stress must have been put on the catheter during withdrawal, irrespective of the material properties. The catheter should slip out easily from the spine, so how was stress induced? The answer came by questioning a hospital consultant in the obstetrics department, and finally, by direct examination of an intact catheter from a recent epidural. The end of the catheter was deformed, mainly due to pressure from the surrounding tissues during withdrawal, and such permanent strain was commonly observed following delivery of the baby (Figure 97). It would clearly require considerable stress to remove the tip of the catheter, so the problem was resolved.
Construct a diagram showing the likely sequence of events, starting with the polymer granules and ending with the failure of the catheter. Indicate on the diagram the evidence on which each event is based. Indicate the main areas of uncertainty in the sequence, and what action could have been taken to reduce that uncertainty.
Based on the information supplied in the case study, Figure 107 indicates the main sequence of events leading to the final failure – some of which is inevitably conjecture. Down the right side of Figure 107, pieces of evidence suggest or corroborate the event in the box.
The main areas of uncertainty are below. In brackets, are some actions that would reduce the uncertainty.
Possible photo-oxidation after manufacture. There is no direct evidence of exposure to sunlight. (Take witness statements from staff in Ireland; plus do further FTIR microscopic examination of the failed catheter to improve the resolution of spectra.)
Possible further gamma radiation degradation of catheter. (Extend literature search for any information on gamma radiation effects on copolymers, nylons, or poly ethers.)
Interpretation of fracture surface features. (Clean sample and re-examine; plus do FTIR microscopy on failed end.)
The case was settled by mutual agreement between the three parties the weekend before the trial was due to start in the High Court, with substantial savings in costs. The hospital paid its own costs, and the manufacturer of the catheter paid damages to Mrs K.
The case highlights many of the problems faced by investigators not just in medical cases, but also more generally. Firstly, hasty attempts to study the surviving evidence can often lead to misleading conclusions. In the catheter case, the initial examination of the failed catheter tip with a hand lens had produced a mistaken inference that the device had failed by being cut, despite the evidence of brittle behaviour elsewhere on the tubing. It is perhaps a natural human response to a request for assistance to provide an instant explanation, but is no substitute for deeper, more penetrating analysis.
Secondly, the sample had not been preserved at all well in the period between the accident and the deeper investigation the device demanded. The failed catheter tip was badly contaminated with debris – presumably from a solicitor's office – and no attempt had been made to store the device under dark conditions so that no deterioration could occur. It is likely that further damage had been created by exposure to light in the four years between investigations. Part of the problem lies in the long time it can take to bring court actions, and the lack of expertise among those holding key specimens upon which a case might hinge.
A third problem lies in the new materials and process methods that have been introduced so rapidly there has been no time to appreciate some of the subtleties of property variation with time, degradation, interaction with other substances, and so on. The catheter was made from a new kind of thermoplastic elastomer, where knowledge of its sensitivity to its environment was known and published – like the French paper in Figure 104. It is uncertain, however, whether or not that information was available to, or had been acted upon by the quality control staff at the device manufacturers.
The lack of communication of key information on material properties also, paradoxically, extends to materials that have been available for much longer periods. Silicone rubber has been used in many medical devices, but the design of some has led to serious and extensive failures, that of breast implants being a notable example (Box 24). This elastomer is notoriously sensitive to fatigue, and is in any case not a tough material, showing poor mechanical behaviour. Designers using the material for products that will experience stress in service, must recognize this fact and design in ample fail-safe features to withstand such stresses.
Extensive reporting of breast implant failures started to occur in the USA in the late 1980s. Most of the failures have occurred from permanent implants made from silicone rubber filled with silicone gel. However, saline implants are often used after the mastectomy operations that are done to treat breast cancer. They are expanded incrementally by repeated additions of saline solution to a silicone pouch, so that the tissue slowly expands to give a space ready for a permanent implant. Such tissue expanders have also failed, especially towards the end of the filling stages, when the load is greatest.
One such failed tissue expander is shown in Figure 108. The fracture occurred at the joint between the catheter to the upper dome, used for injecting saline, and the bag itself.
The material of both catheter and bag is crosslinked silicone rubber, an inert material, so good for implants, but rather weak mechanically – so bad for stressed implants. The junction failed suddenly and caused considerable distress to the user. So what caused failure? Examination of the failed surface in the ESEM showed a crack had grown in one event from one side of the junction, where there was a sharp corner between the catheter and the bag. See the fracture surface map in (Figure 109).
There was a single origin at the junction, but there were also several areas of microcracks elsewhere in the junction. Presumably, the stress concentration at the sharp corner led to rapid growth of one of the cracks there. But what caused the microcracks to develop in the first place?
It was suggested that one possible cause might be the adhesive used to bond the two parts together, because if the density of crosslinks is too high, the adhesive will be both stiffer and less tough than normal. Normal stresses from body movement may therefore have caused minute cracks to develop early in its life, so that when the load from the bag was greatest, the largest one at the junction became critical.
As silicone rubber is known to be weak when tensioned or fatigued, care is needed in the design of critically stressed areas such as junctions. Any stress raisers should be ameliorated as far as is possible, and the device provided with substantial redundancy, or fail-safe features.
Two of the three cases studied in this section under the heading of piping problems were settled before trial, a common result for product liability claims. An approaching trial sharpens the issues, and expert meetings focus on the shrinking areas of disagreement between the experts. Further key tests are performed quickly, attempting to clarify and define those issues more closely.
One area of uncertainty in many medical cases is the question of stresses in the body. It goes beyond the problem of the storage tank in Section 6.2, where the contents were of known mass, and the wall loads could be calculated easily. True, the mass of a breast implant contents are fixed and known, but there are extra loads imposed by the body, which are not known in detail and can be difficult to determine exactly. A woman with an implant will impose extra loads when exercising, or when moving the upper part of her body, or even when just sleeping. For similar reasons, a designer should provide the greatest possible safety margin when specifying catheter diameters, and the amelioration of known stress raisers. Medical failures are a key part of the learning process, but should not displace well known engineering principles.
Surgeons who help with device development go to great effort to inform colleagues of their own failures, perhaps the best example being heart surgery. If mistakes happen, the life of the patient may be at stake, so there is a clear incentive to publicise failure in specialist learned journals, and learn the lessons for product design. For example, there have been many failures of stents during angioplasty operations in arteries, but there has been a feedback loop to stent designers to improve the materials of construction, and also to maximise the balloon strength. Indeed, heart surgeons have a special kit of tools designed specifically to retrieve the remains of stents should they fail mechanically in the body.
No doubt the importance of adequate quality control testing and sampling has been appreciated by the extrusion company that made the catheter, and the medical manufacturer who supplied it to the hospital. The requirements of regulatory bodies (such as the MDA in the UK) are stringent, and rogue or maverick products should not enter the supply chain. The catheter was on the edge of failure, but still possessed enough residual ductility to allow its use before the failure occurred. In the event, the medical product company was taken over by another during the time it took for the action to come to trial in 1995, due liability being assumed by the original insurers.
Describe the key issues where the experts in the catheter case agreed, and those where they disagreed. Draft a one page statement of areas of agreement and disagreement to inform the legal teams in the case, and the court. Assume the experts meet on the Friday before the start of the full trial the following Monday. Suggest why the case is settled over the weekend, just before trial.
The three experts agreed about the following matters.
The catheter cracked in the early stages of the childbirth, and failed only at the last stage of the procedure.
The distal tip of the catheter was left in the spinal column of Mrs K.
The catheter was made by extrusion in the Republic of Ireland and then modified by cutting to length, one end sealed and three holes cut near the tip for drug infusion.
The tip of the catheter failed across the width of the tube, across the proximal hole. The failed surface showed traces of ductility, together with flat zones.
Mechanical testing and comparison of new and failed catheter tubing showed that the new product was tough and ductile, but some parts of the failed catheter were brittle.
FTIR, DSC and NMR results generally showed the material of the failed polymer tubing to be normal, although one spectrum was abnormal.
The main areas where the experts disagreed were as follows.
The cause of failure of the catheter at the end of the birth. The claimant's expert maintained it had been cut by the anesthetist.
The degradation of the catheter during manufacture. The manufacturer's expert probably believed either that it may have degraded by exposure at the hospital, or alternatively, that it had been cut by the anesthetist.
The expert acting on behalf of the hospital believed that, although the direct evidence was thin (one FTIR trace), the admitted brittleness of the catheter during the epidural could be explained by degradation. In addition, the failure surface showed brittleness as well as ductility.
The joint statement (if it had been made) might have been constructed as follows, although there are many ways the statement could have been expressed. Needless to say, the exact form of expression and the slight variations in meaning are quite critical to any ensuing trial.
Statement of areas of agreement and disagreement
The experts agree the catheter failed when the tip was pulled from the spinal column of Mrs K, in the final stage of the birth. They agree the catheter probably showed brittle behaviour prior to the final failure. They disagree about the cause of the failure.
The experts agree the failure surface when examined using ESEM, showed traces of ductility and other flat zones. They disagree about the interpretation of the flat zones.
The experts agree that when other parts of the failed catheter were tested mechanically, several brittle fractures were observed. They disagree about the interpretation of the brittle state of the catheter.
The experts agree that chemical analysis of the polymer of which the catheter was composed showed, in one example, traces of abnormality. However, they disagree about the implications for the strength of the material.
The case was settled because the strength of the evidence for brittleness in the catheter before the tip finally broke during withdrawal was strong, and could be explained by some of the chemical evidence. That explanation was supported by independent evidence from the scientific literature – the French paper. In addition, the fracture surface itself did not show the parallel marks characteristic of cutting by the Tuohy needle. And it was clear from the direct evidence of medical staff, that the tip was stressed during the final stages of the operation. It was likely, on the balance of all the evidence, that the material had been degraded before the operation started. The chances of degradation during manufacture were greater than at the hospital. Liability therefore lay with the manufacturer rather than the hospital. A trial would have sharpened the arguments, and would have tested the witness evidence, but it is unlikely that a credible alternative explanation would have emerged.
This unit has examined a series of case studies involving the failure of various devices through design faults, material degradation, or poor processing. In the case of the ladder, it failed because the user failed to realise the importance of using the ladder at a high angle of repose. Many of the problems were resolved when convincing explanations could be provided by expert evidence. A small number resulted in actions brought by the aggrieved or injured parties, but most of the cases considered in this unit were resolved without the need for trial. This is typical for most cases involving product liability. The majority are settled before trial commences. Indeed, the approach of trial sharpens the arguments, and they devolve onto a small number of issues, or even one issue that becomes the key to the dispute. Discovery is part of the process where new information comes to light, and may provide clues to an explanation of the failure that led to the dispute.
A few disputes do proceed to trial however, especially where serious injuries or deaths are involved following a material or design failure, such as the fuel line case in Ireland. The claims are usually large, and it is justifiable to pursue a deeper investigation into the causes of the failure. Alternatively, the claim may be relatively low, but there could be a principle of law at issue, so the case proceeds to trial for the determination of that issue.
This unit started by looking at simple examples of product failure, and then looked at case studies that are more complex where polymer components or products failed. In the restricted space available in one block of the course, it is impossible to examine all possible failure modes of polymers, which are themselves just one group of materials available to design engineers. Before moving on to learn about metal failures in the next block, it is useful to take a wider view of polymer failures.
Table 6 shows failure modes classified by mechanism, and it indicates the range of ways polymer products can lose their integrity. The most common failure modes are mechanical in origin, for the simple reason most products will bear some load in service – even if it is just its own weight. And many of the other mechanisms will probably lead to mechanical failure in the chain of causation. In this respect, chemical attack will lower strength, and an obvious symptom will be cracking, excessive wear, or some other mechanical effect.
|mechanical||fracture (brittle or ductile or mixed mode)|
|creep rupture (static fatigue)|
|fatigue (slow crack growth from cyclic loading)|
|impact (rapid fracture)|
|wear, fretting (abrasion between surfaces)|
|fire, slow combustion|
|chemical||oxidation, ozone attack|
|chlorinolysis (attack by chlorine), fluorinolysis|
|hydrolysis (by water, acid or alkali)|
|stress corrosion cracking (SCC)|
|other chemical interaction (e.g. ZnCl2, ethylene oxide)|
|optical||radiactive attack (e.g. UV attack)|
|ionizing radiation (e.g. gamma radiation attack)|
|electrostatic build-up, sparking|
Table 7 lists causes of failure from raw material processing to consumer use, and it relates the basic mechanisms to the actual ways products can fail. Table 6 would be the starting point of scientific analysis, but Table 7 would be the route to identifying the detailed reasons for a failure. The way an investigator tackles a specific case will also follow the same path, because they will usually be provided with a failed product, which, when analysed, reveals a specific mechanism of failure. When the wider facts of the problem are known, the investigator can relate the circumstances of failure to the specific mechanism already identified.
Taking the major cases studied in Sections 4, 5, 6 and 7 of this unit, identify the main mechanisms and failure modes described in Table 6. Then specify the main stage at which failure probably occurred, by reference to Table 7. Which case studies showed one mechanism on initial inspection, but showed a different mechanism on further inspection? What major lessons should be drawn from your conclusions? Indicate the role of forensic methods in each case study.
The major case studies examined in Sections 4, 5, 6 and 7 were:
a ladder accident;
a failed composite car radiator;
a fractured storage tank;
a broken acetal fitting on a rising main;
failed flexible fuel pipes;
a failed catheter.
Apart from the ladder accident, they were all associated with fluid containment in one way or another, so failure was quickly detected by catastrophic consequences. The catheter was somewhat different in that failure of the tip was observed without loss of fluid, although the failure at the proximal end possibly caused loss of the infused drug solution.
The main failure modes are as follows.
The ladder accident was primarily a user failure – failure to heed warnings, failure to follow advice – who leant the ladder at too low an angle against the wall of his house. All new ladders have a warning notice posted on the stile of the ladder at head height. The fracture of the tip was not a cause of the accident, but rather the result of the ladder slipping down the wall and striking a ledge above the patio doors.
The radiator reservoir failure was a manufacturing problem, probably caused by a machine operator wrongly classifying the faulty moulding as a good component. The symptoms of failure included brittle fracture of the composite material, aggravated by frozen-in strains as shown by distortion of the moulding – from the low-temperature conditions at start-up of the machine. The failure was mechanical but stimulated by high operating temperatures, and initiated by weld lines and cold slugs in the wall of the product.
The tank failure was a design problem, where the walls had been built like a barrel rather than like a dam, and didn't resist hydrostatic pressure. The immediate symptom of failure was cracking in the centre of the lower single panel wall, caused by creep rupture at each of the four loading cycles. Initiation occurred from a pinhole defect from welding, probably the final weld made at the manufacturer.
The acetal fitting probably failed by chlorine attack at a pre-existing defect or defects in the inner surface adjacent to the screw threads. The mechanism was probably creep rupture from the normal closure stresses. There were thus manufacturing defects that allowed a stress corrosion cracking mechanism to form the cracks.
The fuel pipe failures were caused by two mechanisms: ozone cracking of the NBR rubber pipes – for which a recall was made – and rapid abrasive wear of the nylon 11 fuel return pipes abrading against hot engine manifolds. They were design faults, the materials were inadequate for their intended function.
The catheter failure was probably caused by a combination of attack by the UV component of sunlight, followed by gamma ray radiation before supply to the hospital. The immediate cause of failure was separation of the catheter near the proximal hole in the tip during the last stages of the epidural injection. It was therefore a manufacturing problem.
Several of the cases considered here initially involved causes that were either not relevant to the reason for failure, or more distantly related than first thought. The broken ladder tip wrongly led the user to suggest the moulding was faulty and therefore caused the accident. The immediate cause of the storage tank failure was crack growth in a weld, but poor welding was not the cause of the incident, rather, it was the basic design of the tank that caused the failure. One investigator suggested the fracture of the acetal fitting was caused by high tightening loads, but deeper analysis showed the failure was actually caused by SCC of a faulty moulding. Initial inspection of the tip of the catheter attributed the cause to the anesthetist withdrawing the catheter wrongly, and so cutting the tip off. In fact, the catheter had been weakened by degradation and showed partly brittle properties, so it could have failed easily during withdrawal.
These examples demonstrate that many failures require deeper thought, and the appropriate analytical method for resolution. Corroboration using independent methods is usually required, but that may not always be possible if the sample is very small, or if it requires independent examination by another expert. The ladder problem yielded to a full reconstruction of events, while the radiator reservoir needed microscopic inspection of the fracture surface to show cold slugs on the weld lines. The tank failure was solved when inspection of the standard showed it had been constructed wrongly. The examination of the fracture surface revealed only that the failure had occurred at the weakest point in the product. The critical method in the acetal fitting failure was finally EDAX analysis of the surface to show the presence of chlorine, corroborating the results of the USA litigation. The fuel pipe failures were solved essentially by simple observation of the failure mode and remedial action in the UK – but not in Ireland. The turning point in the catheter case came from a single, uncorroborated, FTIR microscope spectrum. However, the result confirmed what had already been suggested by the brittle features present in the catheter during the birth, the partly brittle nature of the fracture surface, and the poor mechanical properties of the rest of the catheter when tested later.
Delay can result from a product failure that on first analysis, turns out to be more complex in origin than it first seemed. This is partly why disputes can last several years, while key samples are examined by experts appointed by the different parties to those disputes. Delay is also built into the legal system, where each party prepares its position after allegations are formalised into writs and pleadings. There may follow yet further delays, before one of the parties initiates further deadlines by applying to the court for discovery, or for further directions.
Experts who are instructed early in a dispute tend to have some advantage in being able to prepare well in advance of further proceedings, and can often use the time delays built into the system for further research exploring the key issues at a deeper level. This might include extensive literature searches, for example, to discover examples of the kind of failure reported on in the initial report.
Reports frequently evolve as new evidence becomes available – detailed witness statements, perhaps – and the pace of research quickens rapidly as trial dates are fixed by the court. Although most product liability cases are resolved before trial, all reports are prepared in the full knowledge they will be exposed to the full rigour of open trial.
The most demanding analysis occurs during cross-examination, when the barrister for the opposing side will probe all matters relevant to the case. In modern court practice, cross-examination will often be continued by the judge, who is usually a former barrister well versed in critical analysis. The expert can expect a thorough and searching analysis of her research. Some of the aims of cross-examination are to test:
the credibility of the expert witness;
the reliability of the evidence examined by the witness;
the reliability of the arguments used by the expert to suggest the failure mode;
any ambiguities or contradictions within the expert report;
any conflicts between the expert evidence for the various parties in the dispute.
In most trials, the barrister is not normally technically qualified, so will be advised by the expert acting on behalf of the opponents. Nowadays, courts expect expert witnesses to report directly to the court when preparing final submissions, so that the expert has a duty to provide neutral, unbiased opinions.
In those courts that deal frequently with technical disputes, the situation is different, because the barristers and judge often have some kind of technical background, so can appreciate scientific or engineering arguments. The Technology Court, the Commercial Court, and the Patents Courts are usually fully equipped to deal with complicated technical issues.
Whatever knowledge a barrister can bring to a case, he will explore the logic of an opinion, so the expert must be prepared to be challenged on the arguments he or she uses to interpret the evidence. It also calls for a balanced attitude to conflicting pieces of evidence or opinions, and a fair interpretation of the facts established by the court. For instance, if cross-examination shows a piece of witness evidence that the expert has previously relied upon to be false, the expert must be ready to modify her opinion to allow for the new facts. This is not unusual as a case develops at trial, because cross-examination is a vital tool in revealing new and critical issues in a dispute.
One new development that has affected the way experts approach a case, has been initiated by the courts themselves. Broadly known as the Woolf reforms, they are an attempt to speed up litigation – so reducing costs – by giving wider powers to the court to control the way a case develops. Cases are divided into several tracks, depending on the value of the claims.
Small claims track: £5000 or less in the Small Claims Court.
Fast track: £5000 to £50 000 in the County Court.
Multi-track: £50 000 plus in the High or County Court.
However, if a case involves any complexity or is a test case, a High Court trial will probably be necessary.
For expert witnesses, there are several extra responsibilities: an overriding duty to the court; giving a statement of truth with the report; and a disclosure of previous experience, normally done by including a detailed curriculum vitae with the report. Many of the new responsibilities have in fact been used by many, if not most, experts for many years, and have become embedded in current case law (as Box 25 discusses). The effects of the Woolf reforms are still working their way through the legal system, but will no doubt have many effects in the years to come.
The role of the expert in trials is important for several reasons. The main reason for using expert evidence is to provide independent views concerning the facts at issue in a dispute. Unlike witnesses of fact, the expert witness can offer an opinion in disputed areas. This is normally excluded for witnesses of fact, who cannot for example say what he thinks about a given issue.
In the criminal courts, the forensic evidence of fingerprints, or blood type, or DNA matching is often critical, especially where there is no direct eye-witness evidence or when the witness evidence is conflicting – a common event. In civil trials for compensation after a serious accident, expert evidence is important for establishing, for example, whether a particular machine or component was faulty and caused the accident.
The evidence given by experts is a duty directly to the court – re-emphasised in the recent Woolf reforms – rather than to the party who hires him or her, and pays for their time. Problems frequently arise because of the confusion that some experts may have over their duty to the court, and has raised the spectre of the hired gun. If such experts show bias towards their client, they can expect a tough time in cross-examination on the witness stand.
In a case from the 1980s, The Ikarian Reefer, the judge (Cresswell J) described the various responsibilities of experts. Their evidence should be:
independent of the litigation;
objective and unbiased. Expert witnesses should:
state facts and assumptions on which the opinion is based, together with any facts that could detract from the conclusions;
give evidence within their expertise;
only give opinions provisionally if not all the facts are known;
ensure any changes in opinion after the exchange of reports are sent to all sides in the dispute;
ensure all photographs, plans, survey reports, and other documents referred to in the report, are included in the exchange before trial.
These rules have now been incorporated into the Supreme Court rules under the Woolf reforms. They are accompanied by recommendations on the meetings of experts before trial to draw up a schedule of areas of agreement and disagreement; details of experiments and who performed them, with their qualifications; in addition to a statement of truth that must accompany every report. The new rules also emphasise that the expert witness reports directly to the court and is responsible to the court. A typical statement of truth is shown in Paper 7.
Click on the 'View document' link below to read Paper 7.
The cases described so far in this block have had drastic effects on those concerned, which is why disputes arose over their causes. If they were design failures, they would be widespread, although individual faults or misuse can affect just one product, as the Titanic disaster illustrates (Table 8). As decided in the High Court a few years after 1912, the sinking was caused by travelling too fast into a known ice field.
The Senghenydd disaster of 1913 caused the worst casualties the UK has ever seen in a coal mine – 439 died. This was a disaster caused by a methane explosion that then ignited a coal dust explosion, which then travelled throughout the workings. Other more recent colliery disasters include that at Cresswell Crags in Derbyshire, when a rubber conveyor belt caught fire. The result was a change to regulations so that PVC belts were designed for colliery use. It is more resistant to fire than natural rubber. Other disasters involving polymeric materials are highlighted in Table 8.
Such disasters are, of course, thoroughly and exhaustively investigated to determine the cause or causes. Just like most of the case studies examined so far, important lessons for designers will be the normal outcome. With major disasters, there is naturally greater public interest, and changes to existing legislation will often be the result.
|disasters||marine||SS Titanic sinking (1912)||travelling too fast into known ice field; possibly low quality steel|
|mining||Senghenydd explosion (1913)||gas and coal dust explosion|
|Cresswell Crags (1954)||fire from natural rubber conveyor belt|
|Markham colliery (1973)||fatigue of steel brake rod|
|structural||Tay bridge (1879)||poor design; low quality cast iron with many casting defects|
|Summerland fire (1965)||poor design of building; use of PMMA cladding|
|Alexander Kielland oil rig (1980)||fatigue of support structure|
|aerospace||Comet crashes (1955)||fatigue from porthole corners|
|Challenger space shuttle (1988)||design of rocket casing; use of low resilience Viton o-ring|
|nuclear||Chernobyl (1984)||systems failure|
|business||Kodak infringement (1986)||copying of instant photography patent|
|BP infringement (1998)||copying of acetic acid resin patent|
|serious accidents||automotive||Ford Pinto fires (1967)||poor design of petrol tank|
|Fiat Mirafiori fires (1978)||polymer fuel lines and electrical defects|
|widespread failures||structural||miners’ lamps (1974)||poor design, manufacture of product in polycarbonate|
|plumbing failures (1988)||polybutene, acetal in hot water plumbing pipes and fittings|
|accidents||structural||storage tanks (1995) welded tanks (1998)||poor design of polypropylene tanks|
Summarise the main design or quality lessons of the major case studies considered in Sections 4, 5, 6 and 7 of this block. Include prime materials producers, manufacturers, designers, and consumers or users in your discussion.
The main lessons of the case studies of Sections 4, 5, 6 and 7 are as follows.
Ladder accident. The case highlighted poor consumer knowledge in the use of ladders, despite written advice on the product. One design lesson might be to improve ladder design so that ladder stability is not dependent on a single angle of repose. It was suggested that rotating feet are one design option that ameliorates the single angle restriction. Devices that can be attached to the lower part of a ladder are also available, but still require accurate location to be effective. Another lesson for consumers might be that they need some education in ladder use – this might come from government agencies.
Radiator reservoir. The defective radiator reservoir was a maverick product, the lesson being primarily one of improved quality control for the manufacturer and moulder. Machine operators require on-the-job training to spot non-specification products, so they can prevent mavericks being used in final products.
Storage tank. Storage tanks made from thermoplastic material must be designed according to best practice, as defined in this case by a German standard, DVS 2205. Welding should also be improved by pre-forming walls to eliminate frozen-in stress. An additional lesson would be to improve bund wall design so that high leaks are contained safely.
Acetal fitting. The main lesson of the acetal fitting failure must be to reduce the use of acetal with water that might contain chlorine. The advice could be given by the prime material manufacturers in their technical brochures. In addition, injection moulders should tighten quality procedures to eliminate the supply of maverick products into the market. A subsidiary lesson of the USA experience is that manufacturers must listen to their laboratory staff, and make sensible decisions about the widespread use of specific materials in an environment known to be harmful to the material in question.
Car fuel pipes. The use of materials appropriate for their intended purpose is the message for car designers. Fuel pipes are one of the simplest of components but are safety-critical, so the component requires care in choice and specification. The consequences of a poor choice or poor fitment could be catastrophic for the users. Car mechanics should also be aware of the two failure modes possible in NBR and nylon 11.
Catheter. Materials for medical use must be tested and monitored to a higher level than those used in non-medical products because there must be no leaching of chemicals into the body. At the same time, greater care is needed by suppliers and assemblers to ensure sensitive materials are not exposed to sunlight at any time – further exposure to gamma radiation could continue a degradation process started by UV radiation.
You should now watch the video sequence 'An eye for an eye?'. Is based on a fictional court case taken from real case, which involved failure of materials.
Click on the 'View document' link below to read 'Video notes'.
Click on the link to view the video.
Click on the link to view the video.
Click on the link to view the video.
This free course provided an introduction to studying Engineering. It took you through a series of exercises designed to develop your approach to study and learning at a distance, and helped to improve your confidence as an independent learner.
The content acknowledged below is Proprietary (see terms and conditions) and is used under licence.
Grateful acknowledgement is made to the following sources for permission to reproduce material in this unit:
Figures 7, 8 and 10: Kane, V. E. (1989), Defect prevention: use of simple statistical tools, © 1989 Marcel Dekker, Inc. All Rights Reserved; Figures 40 and 71: Neuber, H. (1958), Kerbspannun gslehre, 2nd edn., Springer, Berlin. Translation, Theory of Notch Stresses, Office of Technical Services, Dept. of Commerce, Wash. D.C. 1961; Figure 61: © Open University/British Ladder Manufacturing Association; Figure 63: Courtesy of Jaguar Cars Limited; Figure 74: ‘Safety data for sodium hydroxide’ The Physical and Theoretical Chemistry Laboratory, Oxford University; Figure 76: Inglis, C.E., (1913) ‘Stresses in a Plate Due to the Presence of Cracks and Sharp Corners’ and Kolosoff, G., (1910), Dissertation, St. Petersburg; Figure 90: Schnell Publishing Company (1991), ‘Plastic Pipe is Expensive for Industry’, Chemical Marketing Reporter, 18 March 1991, © 1991 Schnell Publishing Company; Figure 95: Telegraph Group Ltd. (1997) ‘Fiat recall mirafioris for check’, Daily Telegraph, 12 July 1997 © 1997 Telegraph Group Ltd., Courtesy of Telegraph Group Ltd; Figure 96: The Irish Times (1994), ‘Court asked to strike out Fiatfirms' defence in car fire case’, The Irish Times, 12 October 1994, © 1994 The Irish Times, Courtesy of The Irish Times; Figure 100: Donald, A.M. and Kitching S. (1998), ‘Beam damage of polypropylene in the environmental scanning electron microscope: and FTIR study’, Journal of Microscopy, vol. 190, Pt.3, June 1998, Blackwell Science Ltd, © 1998 The Royal Microscopical Society.
Resource E: All photographs reproduced with the permission of the copyright holder, Peter Lewis
Paper 1 (pp. 4–24): A. Finlay, and P.R. Lewis (1990), ‘Total quality manufacture of stationery products with GMEA methods’, Seminar Proceedings vol. 2, London (1992)
Paper 2 (pp. 26–48): ‘The Open University thanks the International Electrotechnical Commission (IEC) for permission to use the following material. All extracts are copyright © IEC, Geneva, Switzerland. All rights reserved. Further information on the IEC, its publications and its role is available from www.iec.ch. IEC takes no responsibility for and will not assume liability for damages resulting from the reader’s misinterpretation of the referenced material due to its placement and context in this publication. The material is reproduced or rewritten with their permission’
Paper 3 (pp. 50–64): This article was published in Engineering Failure Analysis, vol. 6 (1999), Lewis, P.R., ‘Premature fracture of a composite nylon radiator’, 1999, © Pergamon Press, 1999. With permission from Elsevier Science
Paper 4 (pp. 66–83): This article was published in Engineering Failure Analysis, vol. 6 (1999), Lewis, P.R. and Wiedmann, G.W., ‘Catastrophic failure of a polypropylene tank Part I: primary investigation’, 1999, © Pergamon Press, 1999. With permission from Elsevier Science
Paper 5 (pp. 86–103): This article was published in Engineering Failure Analysis, vol. 6 (1999), Lewis, P.R. and Wiedmann, G.W., ‘Catastrophic failure of a polypropylene tank Part II: comparison of the DVS 2205 code of practice and the design of the failed tank’, 1999, © Pergamon Press, 1999. With permission from Elsevier Science
Paper 6 (pp. 106–110): © Peter Lewis; pp. 106–107: Lewis, P.R. (1999) ‘Degradation of an acetal plumbing fitting by chlorine’, © 1999 ANTEC/Peter Lewis.
All other materials included in this unit are derived from content originated at the Open University.
Don't miss out:
If reading this text has inspired you to learn more, you may be interested in joining the millions of people who discover our free learning resources and qualifications by visiting The Open University - www.open.edu/ openlearn/ free-courses