Forensic engineering has a much lower profile than forensic science: its work is normally associated with civil courts and with many cases not appearing before any court at all. The current strands of forensic engineering can be traced from company laboratories that traditionally have dealt with in-service failures, fire investigators and ad hoc background investigations into high profile accidents, such as the collapse of the Tay Bridge.
In terms of today's forensic engineering, there is an increased emphasis on investigating the cause of failure of consumer items. This is because firms are being sued more frequently about allegedly defective products. There is also a continuing need for the investigation of fires, explosions, air and rail crashes and other important accidents or possible crimes.
This section looks at the methods employed today by forensic engineers in their investigation of accidents. Dr Michael Edwards and Dr Peter Lewis provide an overview of the subject, including case studies of the Challenger, Concorde and Hatfield disasters.
The principal methods used in early forensic engineering included:
- the careful analysis of the service record of the component
- the review of the loads carried
- record of temperatures suffered
- analysis of the microstructure of the material used
- assessment of witness evidence.
This was usually carried out by optical microscopy, X-ray diffraction and chemical analysis.
All of these are still required but the task of the forensic engineer has been aided by the introduction of certain key concepts and methods including:
Fracture mechanics - the analysis of loads applied to bodies containing cracks
Scanning electron microscopes - fracture surfaces can be analysed at magnifications of thousands to determine modes of failure, as well as the chemical composition of key components of the microstructure
Finite element analysis - the determination by numerical mathematics of stresses, temperatures at all positions within a body, rather than the analytical solution of simplified shapes
Computational fluid dynamics - this uses numerical mathematics to determine effects of fluids (gases or solids) on components or structures
Impact dynamics - the use of numerical and analytical codes and models to determine the behaviour of structures when hit by fast-moving projectiles.
Fracture mechanics started with the ideas of Griffith in the 1920s on the fracture of brittle solids such as glass, where he suggested that the fracture strength was inversely proportional to the square root of the length of the largest crack (defect) present. Later in the 1950s, when fracture had been seen in metals without prior plastic (permanent) deformation, the concepts of fracture mechanics were developed.
These can be expressed as:
KIc = QÛfv (_a)
KIc = fracture toughness, a function of the material
Q = geometrical factor, related to crack and component geometry
a = length of crack
By analysis of the crack size causing failure and the material, the stress causing the failure can be estimated and compared to the design stresses expected. With the use of fracture mechanics as part of the design process, especially in high strength materials and in high integrity structures, such as aircraft and oilrigs, it is rare to see instant fast fracture.
More usually cracks grow by fatigue (repeated stresses) or stress-corrosion (synergy between tensile stress and corrodant). This crack growth is a function of stress intensity (KI = QÛv (_a), where Û = stress).
Commercial scanning electron microscopes became available in the mid 1960s. Their great advantages over conventional optical microscopy includes an increased magnification from the 1-2000 of the optical microscope to 40-50000 and an increased depth of focus by a factor of about 300. It is possible to look at fracture surfaces, which will be too rough for optical examination, and examine polished sections at magnifications of more than 1000.
The scanning electron microscope operates by focussing a beam of electrons on a specimen, the point of focus being scanned line by line over the whole specimen. Images can be built up from the secondary electrons, the back-scattered electrons and the X-rays emitted. Within the last ten years, a more advanced type of SEM has become available, known as ESEM (E for environmental). Non-metallic samples do not need a conducting coating, and may be examined up to atmospheric pressures, opening up great opportunities in both forensic science and engineering.
The secondary electron image will produce an image of the shape of the object and therefore will be used on fracture surfaces and on polished and etched sections. Contrast from the back-scattered electrons is a function of atomic number, whilst the X-rays can be used as a detector of the elements present in the sample (EDAX analysis).
This method depends on the dividing up of a body into discrete elements, which may be triangles, squares or other defined shapes, and analysing the response of the body as a whole as the sum of what the response is of each element of the body.
Each element is characterised by the value of a quantity (stress, displacement) at the nodes (corners) of the element and values at points in between were estimated by interpolation. The method was developed in the 1950s for problems in structural engineering, especially in the aeronautical industry.
The first usage of the term "finite element" came in 1960 and the generalisation of the method to solve problems in fields other than structural engineering, such as heat transfer, came in the late 1960s.
Since the calculations were numerical and elements were often small in relation to the size of components or structures, the use of the finite element method was somewhat restricted until cheap powerful computing facilities (hardware and software) became available. Initially, problems tackled by this method were one-dimensional but the developments of the method, especially with increased computing power available, has been extended to two and three dimensions.
Computational fluid dynamics is also based on breaking up a component or structure into a set of control volumes. For each control volume equations will be set up to describe the flow of fluid (liquid or gas).The spread of fire or the cooling of a turbine blade by compressor air, can be modelled.
The technique has been in use since the 1960s, initially in the large aerospace companies.
With the reduced cost of computer hardware and software, the usage of computational fluid dynamics has increased greatly in the last ten years. This has reduced the need for some wind tunnel testing and accelerated the production of solutions to design problems involving fluid flow.
In forensic investigations of fire disasters, such as that at King's Cross and the Piper Alpha platform, the use of computational fluid dynamics to determine the behaviour of fast moving flames has been invaluable and this has owed a lot to developments in the study of aerodynamics.
The modelling of impact events, has been considered vital precisely because experimental work in these areas can tend to be difficult, expensive or, in some cases, illegal. Examples could include the penetration of an aircraft skin by a fragment from a missile, the projection of a non-penetrating projectile against a fuel tank or the passage of a shock wave from a detonating explosive.
Computational methods have become very important in this area. Methods have left the defence and nuclear laboratories over the last 15 years, especially as improvements in computational approaches to finite element analysis and computational fluid dynamics have made the implementation of models to dynamic processes possible.
Although these modern techniques of engineering analysis are very important there is still a place for careful experimentation and reconstruction. A recent High Court case involved a claim that a motorcycle carburettor was prone to icing. An alleged consequence of the icing was that power was lost very quickly, causing the rider to lose control. It was necessary to carry out bench tests to investigate whether icing did occur and later, on a motorcycle whose fuel system had been modified to produce icing, to investigate how the motorcycle behaved when icing started. These latter tests were done in a wind tunnel.
Experiments are also useful in Intellectual property cases, such as demonstrating the action of an RCD when being triggered by an electrical leak or surge. Model building is often crucial, especially if done according to the instructions or specification of a patent. In a recent case before the High Court, such a model of a lawnmower helped to resolve the exact issues at stake in the trial. The patent was found to be invalid, partly because its claims included the prior art shown in the model.
Analysis of witness evidence remains a key duty of experts called by the court. This includes the eye-witnesses of an accident as well as other experts. It covers statements and reports made before trial, as well as direct and cross-examination before the court. Experts frequently have to agree (or disagree) about the key points in a case before trial, so that the judge has a clear view of what separates the parties involved in a dispute. Final resolution of issues often comes only during cross-examination on the stand, but some questions may still remain unanswered. In civil cases, the judge has to decide only on the balance of probability.
If the analysis of major failures is considered, many possible modes of failure would have been reviewed in the safety assessments made when the structure was built. For example, analysis of how the towers of the World Trade Center would behave when hit by an aircraft was considered, the problem being that only smaller airliners were thought of as possible hazards.
Thus, the engineering consultancies carrying out such safety analyses are likely to perform much forensic engineering for courts of inquiry. This means that their engineers are likely to be practising forensic engineering as a part of their duties. This emphasis on subject experience will also mean that staffs of university engineering departments are likely to be involved, again as part of their overall duties.
For the failure of consumer products it may be that local trading standards officers will be the first to pick up a problem and they may refer it to expertise at local universities. Separate failure analysis companies with their own laboratories exist in the United States and are likely to become more common in the UK.
The key to their success will be maintaining a high and varied workload in order to recruit and retain real subject experts. A very specialist branch of forensic engineering is the analysis of patented engineering products, where it is believed that one company has infringed the patent of another. These cases can be very intricate and complex, as well as expensive for both the patent owner and the alleged infringer. With authoritative forensic engineering it is possible that these cases can be settled at an earlier and less expensive stage. With this description of forensic engineering it is likely to be a profession that employs most of its members part-time. They are likely to be chartered engineers, many will have postgraduate degrees in their specialised subject and the more senior could well be Fellows of their professional scientific or engineering institution.