Super Puma Down III

Cracks and Cleanliness

The metallurgist calculated that the markings in the crack had been advancing for between 36 and 100 flying hours before failure. He found this from the marks [behaving almost like tree rings] on the fracture surface. If he could have looked at the fracture on the other piece of the gear [never found], he may have found that more than 100 flying hours could have elapsed.

So, it looks like it was a problem internal to the material of the planet gear that caused it to break up. Rather, that is, than damage introduced externally to the component or outside the mechanism.

What materials does an engineer work with? With what does he design structures? Nowadays, it is most frequently with metals. On some occasions though, he may need to specify a natural material like wood [such as a building structure of internals] or a non-metallic [like fabric for the O2 Arena roof or carbon fibre or plastics for racing car bodies]. For this post however, we will confine ourselves to considering metals.

What is it that is important about metals? Most often it is the raw strength of the material; how great the force is that is required permanently to deform it. Once permanent deformation starts to happen in a component from overload; then we do not know where it will stop. The engineer wants to avoid that. However, there are other strength aspects of metals that may, in certain tasks, trump raw strength such as fatigue strength, creep or toughness. Some metal alloys can be said to specialise in these different capabilities.

A very frequent cause of engineering failures in aircraft is fatigue. That is a failure caused by the repeated application of a stress that is smaller than the ultimate stress that will break a component with a single application. Fatigue is, by definition progressive and usually releases small pieces of metal before final failure. A failure of this type is one of which helicopter designers are extremely wary. It is compressive failure in the bearing race under the balls or rollers, called spalling. It is mainly for such chips that the magnetic chip detectors are seeking.

It helps to understand how to choose a suitable material for a design if we allow ourselves a little over-simplification. For an engineering material, just as for a recipe for something to eat, there are certain defined ingredients and a definition of how they are prepared and treated [cooked?]. For the engineer and metallurgist their chosen material is defined by two aspects. There are the ingredients which are chemical elements [mostly metals but not always]. Then there is the preparation sometimes flattening, [not unlike rolling the pastry for food] or shaping or cutting, and, usually, some form of heating and cooling.

Every material that we use can be had in different grades starting from some form of the raw and passing on up to some form of the extremely pure or refined. From the former to the latter the cost invariably increases. This is not unusual. It is true for a range of fabrics passing from coarse, denim for a pair of jeans to fine silk for a shirt. So it is for metals used in engineering from pig iron to single crystal, nickel alloy for a turbine blade. Such a turbine blade can easily be rated, in terms of the skill with which it is made and its value, as the equivalent of a Fabergé Egg.

Although most workaday components are not made from pig iron; they are at least from mild steel which is. In somewhat simplified terms, iron that had been melted together [alloyed] with around 0.2 % carbon. Most metals however include several components; frequently ten or more. Each has to be carefully controlled; present in the right amount. In the old phrase, to be not too little and not too much. Such alloying components have an important effect on the properties of materials. As we said above, usually it is strength that the engineer looks for but she may be seeking other things such as corrosion resistance or even electrical properties.

These, then, are the materials. What about their preparation?

As we have seen, metal alloys are complex assemblies of crystals. The properties of even a given alloy composition may vary depending on exactly how, on a microscopic scale, the components are arranged. Heating in an oven and cooling carefully is a frequent method for arranging the components of an alloy to provide the grade of strength that the designer decides is necessary.

Another aspect of the preparation is to ensure sufficient cleanliness. The raw materials in the form of ores or scrap often contain traces of other elements or compounds of carbon as impurities. These, after melting, may gather at the crystal boundaries. They may also gather in slightly larger quantities where they force crystals a little apart and are called inclusions. Both will be a weakness or stress concentration. For most components this is only a theoretical weakness. The weakening will either be small or extra material may be designed in.

However, this is cannot be done for some other components that are under greater stresses or where the consequences of failure are too great. For these there is no space for extra material or stress concentrations cannot be allowed to encourage fatigue failure. Aircraft components frequently fall into this category and thus do Main Gear Box planet gears. Because of this, the engineer needs materials that have vanishingly small numbers of inclusions and oxides in the material that she specifies. Clearly, good housekeeping in the production process is essential. But there is one powerful process by which material can be cleaned up. This process is actually two and together they go by the dodgy acronym of VIM/VAR, which stands for Vacuum Induction Melting and Vacuum Arc Re-melting. Good technical data can be found at

Both processes ensure that whilst ever the steel is molten, it is under vacuum to ensure that gases in the atmosphere do not help form brittle compounds that would get mixed in. The feed stock for the VIM part is melted by electrical Induction from coils wrapped around a crucible. It. The end product of this VIM part is a precisely correct alloy composition and also ingots that are of the right size and shape for the next process. The VIM furnace looks like the following.

vim furnace from
Vacuum Induction Melting furnace


However, another important feature of high integrity materials is the small size of the crystals. Normal heat treatment can, if it is extended over too long a time, make crystals grow. If we treat the alloy in such a way as to make the crystals small or ‘refined’, the denser network of grain boundaries makes it stronger. The VAR part does just this.

Here the electrodes that come from the VIM process are again melted in the vacuum, but this time by an electric arc. A diagram of such a furnace is shown below. That is, though, not the clever bit. The secret of making small crystals comes from the way that the newly molten pool of alloy is made to re-freeze or solidify. This can be summarised as “very carefully”. The electrode insertion in the crucible; the rate at which it is withdrawn; the electric current supplied; the rate of cooling water flow; and more; will all be controlled simultaneously. This control will ensure, firstly, that the crystals are carefully formed at the right temperature and, secondly with careful cooling, they will then be frozen at a suitable size without being allowed to grow too large.

schematic simple VAR furnace
Vacuum Arc Re-melting furnace


The end product is an extremely expensive material that has a micro-structure that is so fine, clean and refined that, compared to a batch of the same alloy before treatment, it is a Premier League football pitch compared to a farm meadow.

The material of the Super Puma planet gear as designed was marketed as a material that could be VAR treated or not. If it had been then there should have been no inclusions or oxides to cause a crack. Is this a case where, despite the best efforts, some unpleasant defect still slipped through? The forensic examination of the gear said that the material composition was correct but does not explicitly say that the treatment had been checked. Could it be that this one component had been made from the correct material but which had not had, as the manufacturer’s specification allows, the VAR treatment?

There might be a case for reviewing the design of the planet gear. It is currently one piece doing two jobs; the inner part is a bearing race and the outer part is the gear. As we saw a crack passed from the bearing part and into the gear part and cracked the component asunder. There could be a case, it seems at first sight, to try and squeeze a separate gear shrunk onto a bearing race into the admittedly small space available. Thus the interface between the two would act as a crack stop avoiding this type of failure.

The time that the crack took to propagate to failure whilst only one chip was detected reminds us that although in-service inspection methods must be robust, they must also be sufficiently sensitive. They must detect the deterioration of a critical component before the ability of the component to carry its design load is compromised. In this sad case, those methods did not. A major plank of a safety case is that any damaged material is released from the outside of the components. In this accident, almost certainly, the fatigue failure began and progressed until the last minutes or seconds entirely within the planet gear. Thus it was contained under its surface and away from the oil flow.

All in-service inspection methods must be such that they can always be relied upon to provide advance notice of problems. The other way lies instant catastrophe.

The engineer designers always have to be eternally vigilant about the giant forces that they harness. Those forces allow us to speed about our business. But if they escape their bounds then they revert, in an instant, to savage cruelty and destruction. So it was at six minutes to one o’clock on April 1st 2009 when the drive exploded and the uncontrolled power flung the men on G-REDL from the heights to the depths in less than a minute.


Much of the interpretation and all of the engineering background in these three posts are the responsibility of Isambard’s Lad. Most of the facts and all but three of the images come from the Report on the accident to Aerospatiale (Eurocopter) AS332 L2 Super Puma published by the Air Accident Investigation Branch, Department of Transport, UK.

Engineers’ responsibility

Engineering is one of the three drivers in the advancement of the human race. This blog aims to give to career seekers and also to the general public a taste of how this might be so. They are not well served by the current media. It is an engineer posting: not a ‘scientist’. It describes real professional engineering as it is in the real world usually in the present and occasionally as it was in the recent past.

For comment on this post use the ‘Leave a comment’ link below. For general comment on the blog, contact us at .

Published by

2 thoughts on “Super Puma Down III”

  1. In my view, there are sound reasons why VAR is inadequate for highly stressed safety critical components. Its competitive process, ESR, would (if carried out correctly, which is unfortunately never acheived to my knowledge) has the potential to provide a better quality of defect-free material. However, even ESR can be seen to be unnecessary and expensive. Good casting technology is now sufficiently well developed to provide steels and other engineering metals of astonishing quality but at low cost. I am frustrated that despite my best efforts the steel and Ni-base alloy community simply continue to ignore these hugely beneficial new concepts. The principles have been laid out for years (see my book “Complete Casting Handbook”) and have been demonstrated in foundries all over the world. I guess I am just impatient, and perhaps a poor advocate of my technology. But every helicopter ditching causes me to grieve. I would welcome the opportunity to demonstrate the potential of a new approach.

Leave a Reply

Your email address will not be published. Required fields are marked *