- Metallurgy of Rotorway Secondary Shaft -
Chemistry, Hardness, Strength and Fatigue Data from Failed Shafts
RotorWay Secondary Shaft Material
EPI engaged an independent metallurgy lab to determine the chemical composition of the metal from the failed secondary shafts we examined. This lab is an international company which does analysis work for most of the major aircraft suppliers.
Their analyses showed that the material falls within the specifications for AISI-9310H, with one major exception: the material is short on the specified content of nickel, which is included in an alloy to add toughness and hardenability.
RotorWay claims their shaft is made from AMS 6265 (Aircraft Material Spec), which is a vacuum-arc-remelt processed version of AISI-9310.
The chemical content of AMS metals is more tightly controlled than commercial versions, so it is hard for us to comprehend how several metal specimens, which are allegedly AMS-6265-F, could contain only 75% of the required nickel, which is a critical component of the toughness and fracture-resistance of this material..
The exact chemical analysis of the RotorWay shaft material is as follows:
|Element||AMS-6265-F Specification %||Rotorway Shaft %|
|Nickel||2.95-3.55||2.36 - 2.47 ***|
|trace elements||no spec|
The alloy AISI-9310H is a high-quality, high hardenability, high toughness, case-hardening steel. It is a very low carbon steel, containing only 0.10% (that's the meaning of the "10" in 9310), and what is commonly referred to as "10 points of carbon".
This alloy was specifically designed to be a case-hardening steel for use in applications which require the high surface strength achieved by case-hardening, combined with very high core toughness resulting from the low-carbon, hence low hardness core. It is intended for use in extreme applications, where high contact stress is combined with high shock and impact loading.
In order to achieve the combination of surface hardness and depth of case required for a gear tooth or the inner race of a roller-clutch for example, a substantial amount of carbon must be added to the outer surface of the material, typically by the carburizing process.
In this process, the steel is heated to 1700°F and soaked for several hours at that temperature in an environment which is rich in carbon, followed by quenching and tempering. The carburizing process can produce a surface hardness in 9310 from 59-61 HRc (Hardness, Rockwell-C scale), and case depths as high as 0.060", but that much case depth can require over 8 hours of soak..
Because of the low carbon content of this steel, the as-quenched hardness (core and on-case-hardened areas) falls in the 32-38 HRc range, depending on whether the material is toward the low or high side of the allowable carbon content range.
AISI-9310H is typically used for premium-quality steel gears, frequently in the aircraft industry, and it is the material which we (EPI, Inc.) use for all the gears in our Aircraft Propeller Gearbox products.
However, 9310 is not an appropriate material for a high-fatigue shaft with a highly-loaded, interference-fit bearing. In fact, the relatively soft surface and modest fatigue strength provide an environment ideal for fretting to occur.
Although we have researched a variety of reference works, the only fatigue data on 9310 we found applies to the endurance limits in hertzian stress (contact fatigue, as in bearings and gear teeth) for 9310 in its ultra-high-strength form (carburized and case-hardened to 60 HRc or better).
We found no fatigue data for 9310 in its low-strength (as-quenched, non-carburized, non-case-hardened) condition (32-38 HRc).
We think that lack of data is because non-case-hardened 9310 is generally considered inappropriate for high-strength, high fatigue applications.
In the absence of published fatigue data, we can estimate an endurance limit (EL) value for as-quenched 9310 by using a crude generality in which an EL is estimated at 45% of the UTS (at values below 160,000 psi). That gives an estimated EL of about 65,000 psi. for the shaft material. (Remember that the EL value is a lab number which needs to be adjusted to the specific application.)
There is a risk involved in RotorWay's method of using the fatigue data from an allegedly "similar" alloy (4340) to establish values for as-quenched 9310. These alloys are so different (9310 has similar manganese and silicon, but much less carbon, less moly, more chrome and more nickel than 4340) that the fatigue performance of 9310 would be expected to be quite different from 4340.
Without extensive testing of a large sample of specimens, there is no way to state the fatigue strength of the nickel-deficient RotorWay version of 9310.
However, using the estimated laboratory EL value of 65,000 psi from the preceding paragraphs (for in-spec material), and, using the commonly accepted Marin Method to adjust that laboratory EL for the actual application conditions, we get the following adjustment factors:
|(a) surface finish: 0.87;||(b) size: 0.85;||(c) load: 1.0;|
|(d) temperature (150°F): 1.017;||(e) reliability (99.5%): 0.78.|
Multiplying these factors together produces a composite adjustment factor of 0.587. Applying that factor to the 65,000 psi estimated laboratory EL produces an application-specific EL of about 38,200 psi. NOTE that fretting can reduce this fatigue strength value to well below 20,000 psi.
We tested a total of 30 different points on the broken end of the each failed shaft sample: ten across the fracture face, ten across the finished end, and ten locations at various points on the outside diameter of the shaft stub. Around the fractured area, the hardness of the shaft samples was inconsistent, varying from a low of 33, to an infrequent high of 39, with the median at about 35 HRc.
On the surface which acts as the inner race for the one-way clutch, the surface hardness is consistently between 59 and 60 HRc.
The hardness profile of the RotorWay shaft suggests that it was rough-machined, then masked on all surfaces except the clutch inner-race surface, then carburized, then finish-machined. The clutch inner race surfaces and the adjacent bearing diameters were most likely ground to finished size in order to achieve the required tolerances and surface finish.