Aeroengine Safety

Flight Accident Reconstruction

The goal of reconstructing flight accidents is to determine the (damage) causes through the accident process. In doing so, it is important to be conscious that such complex occurrences usually have several factors that work together to make the accident occur. If observations and facts (e.g. flight recorder data, see Chapter 4.2), damage symptoms, and/or statements from witnesses indicate the accident was caused by an engine failure, the investigation will concentrate on these. This type of damage symptoms include, for example, holes in the engine cowling made by uncontained engine fragments, or external signs of an engine fire.
A satisfactory reconstruction must completely describe the accident process without any contradictions. A reconstruction must be seen as being wrong if the details are not plausible. Reconstructions are made through a combination of different methods. This chapter describes characteristic damage symptoms and the meaning they have for reconstruction of the accident process, especially the accident causes. This information must then be verified by analyzing findings (Chapter 4.1) and operating data (Chapter 4.2), as well as through engine part testing (Chapter 4.4) and analysis of the damage processes (Chapter 4.5).

Fragments and puncturing:

It is not necessarily the case, that sufficiently informative fragments will remain in the engine. With blade failures, especially, the fragments may have so much energy, that they punch through all engine and nacelle walls in their path. The only thing remaining for analysis in this case is the holes in the walls that were punched through and the contact areas between the walls and the fragments.
Naturally, the entire remaining aircraft will be searched for fragments, even in areas where the first signs of a component failure were registered. This is made more difficult by the fact that high-energy fragments, such as those from burst rotors, can fly several kilometers off to the side of the aircraft (Fig. "Fragment location"). Upon impact, they can bury themselves several meters into the earth, depending on the type of ground, which makes finding them extremely improbable. If the fragments escape over water, the probability of finding them is even lower. If all fragments have been lost through the holes and none of the engine part remains, a reconstruction is the only answer.
If fragments of an engine component are created during flight, then clearing up the following points is of primary importance:

• Causally affected components
• Number of fragments and the order in which they were thrown clear
• Size and shape of the fragments
• Location of the fragments in the broken part
• Energy of the fragments.

Important insight can be gained by assembling the engine region and connecting the puncture holes with the plane of the broken components in tangential direction (Fig. "Fragments").

Causally affected components:

This refers to the components from which the fragments originated. In most cases, the identity of these components can be accurately determined. However, if they consist of single parts such as disks and blades, which have separated from one another, differentiating them may be difficult. If the material composition of the components is different, it may be helpful to microanalyze the walls of the holes where the fragments punched through.
If the component that was the cause of the fragments has been determined, actions such as the following present themselves:

• Clearing up the design criteria, life span-determining operating parameters, and affected part groups

• Determining the individual data such as serial numbers, manufacturing data, testing procedures, and test logs
  

• Researching operating experiences, especially if crack initiation has already been reported in comparable engine parts
  

The number of fragments and the order in which they were thrown clear:

The number of fragments can indicate the location of the primary fracture. For example, two symmetrical punched holes are a sign that a disk fracture originated in the hub (Figs. "Housing deformations" and "Damage symptoms of the housing").
The presence of a large number of holes may point to a haircut of the disk blading. With integral disks in small engines (e.g. APUs and helicopter engines), brittle behavior of the disk (e.g. in steels, caused by hydrogen-induced crack initiation or corrosion susceptibility by stress cracking, see Chapter 5.4) can result in many fragments being created.
The order of the blade fractures around the circumference against the direction of rotation is important if it is assumed that the components disintegrated during one rotation. With a little luck, the order in which the blades were flung off can indicate the location of the primary cracks.

Size and shape of the fragments:

The size and shape of fragments can also indicate the probable locations of the cracks. If this realization is combined with the movement of the fragments, it increases the probability that the location of the primary crack initiation will be found.

Location of the fragments in the burst components:

As mentioned above, knowing the order in which the fragments were thrown free, as well as their size, may make a reconstruction of the damage possible.

Energy of the fragments:

The diversion of the fragments from the plane of rotation can indicate the energy of the fragments when they were thrown free. The farther the holes are from the plane of rotation, the less the kinetic energy of the fragments that created them.
A further possible indicator of fragment energy is the thickness of the walls that the fragment penetrated.

Conclusions from unusual observations:

In some cases, in the course of an investigation, effects are observed that cannot easily be made to fit a working hypothesis that seemed plausible until that point. Such cases are depicted schematically in Figs. "Fragment location" and "Injection pipe".
Critical attention should be given especially to damage symptoms that seem astounding. It is important that these symptoms are not ignored, but that an attempt is made to technically and physically understand them and to incorporate them into the damage hypothesis without contradictions. If this cannot be satisfactorily accomplished, then the hypothesis must be regarded as suspect and treated with appropriate care.
This type of anomaly is usually not found until detailed work is being done, such as, for example, scanning electron microscope inspections. In order to interpret damage structures, profound knowledge of materials science, physical effects, and mechanical processes is necessary. An experienced SEM operator with broad knowledge of his subject and the ability to evaluate damaged structures should be on hand for a discussion with the investigator in presence of the SEM. This technical, creative discussion is a requirement for the analysis of the damage process and a proper understanding of the findings.

Examples of physical relationships to be aware of:

• Bernoulli`s Law states that in the region of increasing flow speed, pressure decreases. This means that liquids can be sucked in. These fluids can be fuel, oil, or auxiliary materials, but they may also be molten metal (Figs. "Injection pipe" and "Suction funnel").
• The plastic deformability of metals can decrease at high deformation speeds. This is related to an increase of the yield strength. Therefore, at high stress speeds, such as when a fragment strikes another engine part, materials can behave more brittle than usual (a similar effect has been observed in synthetics, for example).
• The fracture mechanical behavior of certain materials is different in parts with different wall thicknesses. For example, a material that is ductile in thin walls may behave brittle in thick walls.
• The melting points of oxide layers is usually considerably higher than that of the metallic base material. The result of this is that engine part (e.g. turbine rotor blades) material melts, but the oxide layer keeps them in much like a plastic sack, and their geometry can be recognized despite the heavy deformations).
• If brittle materials such as ceramics are subject to impact stress from one side, an elastic wave will travel through the material at the speed characteristic of this material and be reflected by an opposite surface. The reflected impulse overlays with the incoming one, which can create high tensile stress on the side opposite the one where the impact occurred. This can cause spalling and outbreaks of material in areas of the part that were not subject to any direct external impact.
• If metals with low melting points come into contact with nickel alloys, the melted metals may fuse at temperatures that are considerably below those of the nickel alloy.

The reconstruction of an accident uses the results acquired with the aid of the recommendations in Chapters 4.1 and 4.2. Always remember that an incorrect damage reconstruction is highly likely to result in improper corrective measures, which will be expected to result in repeated damage until the truth is found. Wishful thinking and downplaying do not lead to the answer. On the contrary, a third-party expert (e.g. from a regulation authority or the operator) will be able to recognize the actual damage process from the measures taken, even if the investigation results are not very transparently presented.

• Visualize the paths of uncontained fragments!
  

• Try to recreate the investigation findings in iterative steps with computer simulations! It is important that all damage-relevant findings can be simulated without being forced, if possible.
  

• In cases of unclear damage sequences, engine part tests (damage reproductions) have proven to be helpful for understanding the entire damage process for certain hypotheses or for the understanding of individual phases.
  

• Ensure that the damage sequence is free from contradictions! Similar to a court case with evidence, the smallest inconsistencies in the logical sequence, or damage symptoms that the current hypothesis cannot explain, are reason enough to discard the hypothesis in its current form.
  

• Try to discuss with experts who are known to favor a different hypothesis, so that you can hear their arguments!
  

• Always remember that there are connected parties or individuals who have an interest in a certain damage sequence and convincingly and vehemently promote it through a consciously or unconsciously wrong assessment of undoubtedly present damage-promoting factors. Experience has shown that engineers are more likely to search for causes in the material properties, whereas materials experts will be more likely to search for problems during construction or operation, and the maintenance crew may look for flaws in the material. Therefore, it is vital to have healthy skepticism with regard to “facts” that you have not yet verified yourself! Ensure that you verify the statements yourself on-location with original documents and the original hardware!
  

• Determine how many similar cases have already occurred! Assume that many of the earlier cases with similar damage symptoms were not isolated incidents, even if they were classified as such.
  

4.3-1 J. Ott, “Investigators Find Reconstructed Tail of DC-10 Riddled With Damage”, periodical “Aviation Week & Space Technology”, August 7, 1989, page 22.

4.3-2 D. Huges, M.A. Dornheim, “United DC-10 Crashes In Sioux City, Iowa”, periodical “Aviation Week & Space Technology”, July 24, 1989, pages 96, 97.