Appendix C Justification for low likelihood of fatigue

FAILURE

C.1 Purpose of Appendix C

This Appendix provides information related to determining whether a fastener may have a low likelihood of fatigue failure and be verified by similarity to a known fastening system design.

C.2 Justification

A fastening system is considered similar if all of the following are true:

• The design service load spectrum is enveloped by the hypothetical design load spectrum shown in Table 7, Hypothetical Enveloped Applied-Loading Spectrum for Fastener Fatigue, and
• The fastener is made of A-286 alloy, and
• The fastener has UNJ or MJ threads rolled after heat treatment and the limit tensile load multiplied by any factors applicable to fatigue analysis does not exceed 50 percent of the fastener’s allowable ultimate tensile load, or the fastener has UNR threads rolled after heat treatment and the limit tensile load multiplied by any factors applicable to fatigue analysis does not exceed 40 percent of the fastener’s allowable ultimate tensile load, and
• The lowest elastic modulus for the material of any clamped part excluding washers is greater than one-third of the elastic modulus for the bolt material (see discussion in Appendix A.5), and
• The minimum edge distance (measured from center of hole to closest edge of any clamped part other than washers) is at least 1.5 times the bolt diameter (see discussion in Appendix A.5), and
• Requirement TFSR 10 (see section 4.4.3) is satisfied, and
• If threads are in the shear plane, either a dedicated shear transfer device such as a shear pin or dowel pin is present or analysis per the requirements of section 4.4.6 demonstrates no slipping occurs.

Assumptions:

• The bolt is loaded highest by random-vibration response of a 200- Hz mode of vibration.
• The flight structural assembly is proto-flight tested for 1 minute per axis, 3 dB above maximum predicted environment, at three levels of assembly, for a total of 3 minutes per axis.
• The 200-Hz mode is excited equally by all three axes of excitation (conservative assumption).
• Random vibration is at or near maximum predicted environment for 15 seconds during launch.
• The loading spectra for test and launch are based on a Rayleigh distribution.
• Limit load for proto-flight-level random vibration is 3 sigma (three times the root-mean-squared value) and is set equal to 0.5Ptu-allow .
• All loading cycles above 3 sigma from the Rayleigh distribution are included in the cycles at 100 percent limit load.
• The number of cycles at each load increment includes a service life factor of 4.

The similarity criteria is based on fatigue analysis using data acquired in tensile fatigue testing of 160-ksi, A-286, NAS1351N fasteners and a hypothetical loading spectrum intended to be worst case for single-mission spaceflight hardware. The NAS1351N fastener is a socket-head cap screw with a UNR thread form that has a specified root radius and is thus more fatigue resistant than the UN and M thread forms but is not as fatigue resistant as the UNJ and MJ thread forms that have a larger specified root radius. Thus, any conclusions drawn from fatigue analysis using this test data are conservative for A-286 fasteners with rolled UNJ or MJ threads.

The fatigue tests were conducted on two sizes of fasteners, 0.1900-32 UNRF (0.190-in diameter) and 0.3750-24 UNRF (0.375-in diameter). The smaller fasteners were from three lots, each from

a different supplier, whereas the larger diameter fasteners were from a single lot. Three fasteners were tested from each lot for each of eleven combinations of maximum load and stress ratio, R, that is defined as the minimum applied load within a loading cycle divided by the maximum applied load within that cycle. Each fastener was cycled within its given load parameters until rupture or until one million cycles were reached, whichever occurred first.

S-N curves were used to assess fastener fatigue life for a joint and a loading spectrum that together are intended to represent a worst-case scenario for single-mission flight hardware. The joint assessed is the one shown in Figure 10, for which, prior to separation, the bolt’s tensile load increases by a high percentage of applied relative to most joints. The loading spectrum used in the fatigue analysis is shown in Table 7. Each level of applied load is assumed to be fully reversed (tensile and compressive).

To assess the bolt in the subject joint for fatigue life, the applied-loading spectrum is transformed into two bolt-loading spectra. One spectrum is based on the assumption of a high preload, equal to 0.75Ptu-allow , and the other spectrum is based on the assumption of a low preload, equal to 0.5Ptu-allow that is the lowest the preload could be for the separation criteria to be met with a limit load equal to 0.5Ptu-allow . For each level of applied load, the total bolt load is calculated using Eq. 8. For fully reversed applied load, as assumed in this study, Eq. 8 is conservatively assumed to apply for compressive loading as well as for tensile loading, such that the mean load is equal to the preload.

Miner’s rule was used in this study as the failure criterion

(Eq. 87)

where D is the cumulative damage, ni is the number of loading cycles at a given stress level, Ni is the number of cycles to failure at that stress level, and j is the number of different stress levels.

Table 8, Results of Fastener Fatigue Analysis for the Hypothetical Joint and Loading Spectra, shows the results of the fatigue analyses where the stress ratios were calculated using nonlinear finite element analysis. With both loading spectra, the Miner’s rule criterion passes.

UNR thread form. Although the study justifies a limit tensile load as high as 0.5Ptu-allow , a limit tensile load of only 0.4Ptu-allow is allowed for fasteners having the UNR thread form. This apparent discrepancy is intentional as a result of uncertainty, given the limited fatigue data used in the analysis and the assumptions made in the analysis. The UNJ and MJ thread forms have larger root radii than UNR and thus have better fatigue life. Given this fact along with the lack of fatigue data for A-286 fasteners with UNJ or MJ threads, a limit tensile load of 0.5Ptu-allow is allowed for fasteners having UNJ or MJ threads based on the results of the above analysis.

The hardware developer should determine whether their particular loading spectrum is enveloped by the spectrum used in this analysis.

This analysis shows that, for a fastener meeting all of the stipulations for similarity, fatigue under tensile loading is not a concern for the above hypothetical loading spectrum. The test data used in this study are from testing of NAS1351N fasteners that have the UNR thread form.