9. Fires and Explosions

- 9.1 Metal Fires
- 9.2 Oil Fires
- 9.3 Fuel Fires
- 9.4 Dust Explosions
- 9.5 Detecting and Extinguishing Fires

Because engines are powered by fuel combustion, the potential danger of a fire and/or explosion of fuel or its fumes is always present. However, fuel is not the only flammable part of an engine (fuel fire). The lubricating oil, as well, can ignite inside the engine or after a leakage (oil fire). If hydraulic fluid is used near the hot parts of the engine, this can also ignite if leaked. Burning engine parts made of titanium alloys are especially dangerous (titanium fire). If combustible dusts are created, for example from abradables or residue from the engine casing (e.g. magnesium) a dust explosion could occur.

In order to understand the sequence of events leading to, and causes of damage and thus development of specific remedies to prevent future recurrences, it is important to determine whether or not a fire broke out and if so, where. If the damage caused by a fire is extensive, it can make it very difficult to come to firm conclusions. A fire following a crash makes these tasks especially difficult. Volume I, Chapter 4.2, describes traces of soot, burn holes, and other common signs that can indicate a fire began during flight and possibly even shed light on its causes. Contrary to popular belief, the databanks of an engine's digital regulation systems can also be helpful in determining the cause of a fire (Ref. 9.0-1, Example "Recreating fire damage in laboratory").

Evaluating Traces of Overheating

Traces of overheating in damaged engines give clues as to the causes, starting points, and the course of fires (Figure 9.0-1, Example "Recreating fire damage in laboratory"). However, one must be aware of the complexity of these relationships to avoid coming to false conclusions. For example, even if a table of the heat-induced discolorations of the material being inspected is available, the atmosphere during the fire, deposits, or mistaken assumptions regarding time can make conclusions regarding temperature very uncertain. The same is true for time estimations.
The fact that the intensity and duration of a fire influence one another with regard to the amount of heat must be taken into consideration.
The first question to be addressed is, if the overheating was a result of the crash or began during flight. Certain traces can be helpful in this task. Some examples of these may be holes in a casing caused by a titanium fire, traces of fire (soot deposits, etc.) that have been directed by air currents during flight, or rotor disfigurations whose shape has been influenced by centrifugal force.

Place of origin: The place a fire began has certain characteristics that enable one to make conclusions about the direction of the fire's spread. Such a characteristic is the order of burn-related perforations, such as on the combustion chamber casing and bypass ducts. The shape of the burn marks, the order of melted burrs, and hardened drops of melted material are useful indicators. Changes in hardness and structure indicate the direction of the heat source. Deposits, such as oil coke or splattered drops of metal can determine “sources”, i.e. engine parts. If a coating of deposits is present, this can indicate the chronology of the fire, the role of engine parts in the course of the damage, and thus the place the fire began.

Cause of a Fire: The place of origin and the effect on certain surfaces enable conclusions as to the cause of a fire. For example, discoloration of scratches or oxidation of fractures can indicate the time frame and thus the primary damages. The shape and makeup of deposits also help determine the time frame. A fatigue fracture<U> </U>almost always indicates primary damage; a fuel leak after a pipe breakage, for example, that made the necessary stress reversal impossible after start of the fire.

Intensity of a Fire: This means the amount of heat released in a unit of time, and makes conclusions concerning the type and amount of burned materials possible. Titanium fires, for example, reach extreme temperatures and burning drops of the material have an intense heat transfer. Fires of pressurized oil (e.g. after the failure of an oil line), or flames escaping from a burn chamber under high pressure (torching flames), are characterized by far more intense overheating than oil that is simply leaking out.
Changes in hardness and structure across a cross-section of a surface indicate the speed and direction of the heat.

Place of origin and chronology: Analysis of deposits can facilitate conclusions concerning the chronology of the damage. Discolorations and oxidation of fractures along with mechanically damaged surfaces can determine the place of origin. When inspected with regard to the fire's intensity, diffusion processes and the volume of affected cross-sections indicate the duration of the fire.

Fire warnings: There are fire-alarm systems (Ref. 9.5) with sensors in the engine area (e.g. the engine nacelles). False alarms and the activation of fire extinguishing equipment are often due to a malfunction of the system (e.g. problems in the electrical connections). There are other cases, where a faulty wiring or a misreading of the alarm signal by the pilot is suspected. It is thus possible, that on a twin-engine aircraft, the wrong engine is shut down. The assessment of externally visible flames, especially by laymen (e.g. passengers) has been shown to be misleading.

Figure "Traces and causes of overheating": Overheating due to fire can leave distinctive traces, the analysis of which can indicate the cause and course of the damage (see also Ref. 9.0-2):

Discoloration and Oxidation: Especially on metal surfaces that were created by the damage, one can count on the presence of analyzable discolorations. Pronounced oxidation of a fracture in the crack initiation zone precludes overheating at an early point in time and signifies primary damage. The analysis of discolorations requires a great deal of experience and, if possible, simulated tests of surfaces related to the damage under realistic conditions.
Color(e.g. Ti-alloys, yellow, white, and grey) and morphology (rough and porous, loose agglomerates, thick smooth layer, for example) of oxides can enable materially specific conclusions as to the temperatures created.

Penetrating Burns: Holes are usually formed only when there is an intense directed fire; for example, as the result of a concentrated stream or a particle cloud (fuel, oil, burning molten Ti), supported by an expanding air stream. These types of damage occur during flight in running engines. The order of melted burrs and/or the dispersion and shape of melted drops indicate the direction of escaping flames. The same goes for the order of penetrating burns through multiple concentric housings. Additionally, this is a clue as to the origin of the flame.

Directed overheating: Overheated zones are recognizable by changes of the surface of the
parts affected, such as discolorations, blistering or peeling of paints, and/or the dispersion of loose surface particles.
These are created when a flame strikes a surface at a low angle, thus extending the overheating
in the direction of the flame.

Deposits: If deposits were carried by the flame, their composition indicates the flame's origin, and a directed structuring indicates its direction.

Soot and Coke: Lubricating and hydraulic oils as well as fuels can carbonise on hot metal surfaces or leave analysable coatings on metal surfaces when they burn (see Volume 1). These facilitate conclusions as to the composition of the burned material, the chronology, and the conditions of the fire (e.g. the presence of oxygen influences the structure of the deposits).

Plastic Deformations: To be relevant to the cause of the damage, deformations must have been created in flight before the crash. Usually these are creeping deformations resulting from a softening of the material due to overheating. A bulge, for example, shows that a difference in pressure was created in the direction of the bulge. Distorted deformations indicate induced residual stresses in line with local heat expansion in the plastic areas. Conclusions about temperature gradients and the course of the distortions' causes can be made on basis of these.

Structural Changes: Overheating can lead to various changes, depending on the materials involved. Structural components can disappear (e.g. dissolve), change (e.g. grow), or be created. One can assume that a structure whose state is different from specifications was influenced by the overheating. The structural changes give clues as to the temperatures and cooling rates. Infiltrated sintered materials exhibit drop-shaped bleedings with an infiltration phase with a low melting point. Thermosetting superalloys (e.g. Ni-based materials) enter gamma prime at 1070C. If the temperatures lower the materials go out of gamma prime. These excretions can be markedly different from those created in the normal running temperature range. They indicate long time durations or an overheating that took place much earlier. An analysis is more difficult if the overheating was followed by operation at normal temperatures.
Coatings, such as inorganic high-temperature lacquers, diffusion layers, or layers of metals with lower melting points (e.g. silver or aluminium) can undergo changes through diffusion and/or melting above specific temperatures (ex. 9.0-1). Extreme temperatures can cause blistering or a separation of layers (e.g. due to different temperature expansion rates than the base material).

Changes in Hardness: Structural changes usually influence hardness and solidity. Depending on the temperatures reached and the cooling conditions, an increase or decrease of hardness specific to the material involved can occur. To come to conclusions about overheating, it is necessary to know the hardness outside of the specified area (ex. 9.0-1). A change in hardness over a certain distance enables conclusions about the original temperature distribution. If the hardness and structure are changed across a broad cross-section, conclusions about the time of the overheating are possible.

Characteristics of Fractures and Cracks: If there are cracks in a region suspected of having been overheated, a microscopic analysis of the damaged surface (SEM) can give testimony as to the temperature influences (e.g. reaching of the solidus zone/heat crack initiation or creeping), time lapses (e.g. thickness of oxidation layers), and causal relationships (e.g. a fatigue fracture in a fuel line indicates the primary damage).

Figure "Areas where fires must be expected": The typical materials involved in fires and explosions inside of an engine (see above sketch) are fuel, oil, and flammable metals (titanium and magnesium alloys and combustible shavings from flammable abradables coatings e.g. graphite or synthetic coatings). The presence of materials reinforced with carbon fibre in the front compressor blades increases the danger of a dust explosion in the case of serious compressor damage and/or rigorous rubbing.
The bottom sketch shows engine components that are causally involved in the escape of burning or fire-causing materials to the outside of the engine. Most commonly affected are oil- and fuel lines and their attached devices (often with fuel as the medium), hydraulics of the jet-nozzle adjustment, and hot-air lines.
Leakages in these parts occur through breaks and tears (e.g. fatigue), installation and maintenance errors (e.g. forgotten or not properly secured seals (e.g. magnetic seals and oil filling-caps)), not properly fastened bolts, and failing “V-tensioning belts”.
If flanges in the hot part of the engine develop leaks (e.g. through tears or failure of the bolts) or casings are perforated, resulting in escaping flames and hot gases, external fires are likely to start.

Example "Recreating fire damage in laboratory" (Ref. 9.0-1):

Excerpt: “…Laboratory in Ottawa planned to expose aluminium skin panels treated with primer paint to varying temperature levels and study the resulting patterns of discoloration on the paint and metal. Their objective is to determine how much the damage to particular pieces of the … wreckage was caused by exposure to high heat and to match the patterns on the test panels to those on the wreckage.
The technicians are also testing the hardness, ductility and other characteristics of metal wreckage pieces to see how they have changed from production specifications.
…Separately, investigators have completed downloading the nonvolatile memory of the full-authority digital engine controllers from…the three…engines. The data, which recorded fault indications from each engine's systems, is being fed into the investigation's larger effort…“

Comment: the described process can be seen as typical considering the type of problem to be solved. In particular, the recreation of the overheating through defined laboratory experiments and the comparison with material changes in the parts damaged in the accident is normal procedure.

Figure "Causes of engine fires": The causes of internal and external engine fires are very different. The externally visible aspects of the damage: the course of the fire, resultant damages, and the shape and color of the flames and smoke, indicate the causes of the fire. Without claiming universal applicability, the following are descriptions of typical fires:
A dust explosion in the fan (“1”, e.g. after a blade failure with heavy rubbing) or booster (“2”) can result in an explosion-like fire (see Chapter 4) out of the front of the engine. A dust explosion caused by synthetic material shavings is distinguished by a dark red flame and dark cloud of soot.
An violent stall in the compressor can produce a spectacular burst of flame a few meters long out of the engine core (“4”). This explosion is explained by unburned fuel (fumes) behind the combustion chamber, e.g. in the engine`s output. A temporary lack of oxygen prevented combustion in the combustion chamber. Explosions in the bypass region are indicated by escaping flames and smoke in the bypass duct (“3”).
When unburned fuel (e.g. after an aborted attempt to start the engine) in the output area of the engine core ignites, it is called a “tail cone fire”. Due to the low, that is, nonexistent exhaust stream, this type of fire is not directed axially, but is oriented upwards in accordance with the rising of heat.
Titanium fires, on the other hand, usually result in a bundled stream of white flame shooting from the side of the engine in the region of the high-pressure compressor.
Hot gas emissions can be assumed to exist when a fire burns its way out of the engine.
Oil fires are difficult to recognize during normal engine operation. They result in a brief (as long as oil is present), slightly denser soot discharge.
If, due to internal engine damage (e.g. fan blade failure), residue collects or broken fragments are created, a short, intensive shower of sparks in the exhaust stream can be observed.


9.0-1 J.T. McKenna, “Investigators To Reconstruct Flight 111 Nose Section”, periodical : “Aviation Week & Space Technology”, November 9, 1998, Page 100 and 101.

9.0-2 International Civil Aviation Organization, “Manual of Aircraft Accident Investigation”, Fourth Edition-1970, Doc. 6920-AN/855/4, Appendix 12, Page 12-1 to 12-10.

© 2020 ITTM & Axel Rossmann
9/9.txt · Last modified: 2020/06/25 22:43 (external edit)

Page Tools