(meca_9_5)= # Method selection for different items and failure mechanisms Once it is clear what needs to be considered in the {term}`reliability prediction ` (see {numref}`meca_9_4`), the next step is to decide how each part and {term}`failure mechanism ` should be modelled. The present section gives guidance on this method selection task. (meca_9_5_1)= ## Categories and combinations of mechanical failure mechanisms The method selection for mechanical parts reliability modelling should always take basis in a good understanding of the failure mechanisms that are of relevance for the considered item, especially the predominant ones. To simplify the presentation in the following paragraphs, mechanical failure mechanisms are grouped into five distinct categories listed in {numref}`meca-table4-4`. ```{list-table} Categories of mechanical failure mechanisms :name: meca-table4-4 :header-rows: 1 * - Failure mechanism category - Examples * - Distortion - Creep, elastic deformation, plastic deformation, displacement * - Fracture / Fatigue - Ductile / brittle fracture, high-cycle fatigue, low-cycle fatigue * - Mechanical wear - Adhesive wear, abrasive wear, lubricated wear * - Corrosion - Stress-corrosion cracking, galvanic corrosion * - Material degradation - Thermal degradation, radiation damage ``` The probability of failure can for some items be driven by several different failure mechanisms. When using {term}`PoF` methods, the results for the individual mechanisms should be combined to estimate the overall part reliability. The correct way of doing this depends on whether the different failure mechanisms can be considered as independent processes or events. For independent failure mechanisms, it is sufficient to consider one process at once, and to combine the results using normal probability calculus. However, independence between different failure mechanisms cannot always be a valid assumption, e.g. in case of physical interactions between different failure mechanisms processes. {numref}`meca-table4-5` gives an overview for which combinations of failure mechanisms such effects can have to be considered. The table distinguishes interactions between different failure mechanisms affecting the same part (left hand side) and between different failure mechanisms affecting different parts (right hand side). ```{list-table} Possible interactions between different mechanical failure mechanisms :name: meca-table4-5 * - ![table5_4_5](../../pictures/table5_4_5.png) ``` A simple way to account for such interactions is to select the dominating failure mechanism (e.g. the one with the highest probability of failure) and to consider the effect of other failure mechanism(s) on the dominating failure mechanism. When using structural reliability methods, this can be achieved by adapting the basic variable distributions to account for physical effects related to other failure mechanism processes. The assessment then requires the following steps: * Selection of the dominating failure mechanism, e.g. wear of a lubricated bearing. * Identification of other failure mechanisms that can interact with the dominating failure mechanism using {numref}`meca-table4-5`, e.g. distortion of the membrane compressing the bearing. * Identification of physical variables used by the model for the dominating failure mechanism that can be affected by the interaction, e.g. increase of the contact pressure in the bearing. * Adjustment of the identified basic variables distributional models to account for the effect of the interaction, e.g. by considering the increased contact pressure.. Finally, it should be noted that besides physical interactions between failure mechanism, there are other possible reason for statistical dependence, such as common cause effects due to random physical variables (e.g. temperature, loads), influencing different failure mechanisms and/or parts. An example for explicit common cause modelling in a redundant bolted interface with load redistribution can be found in {cite:p}`mec-AO-1-8017-14-NL-KM`. (meca_9_5_2)= ## Selection of methods for part level mechanical reliability prediction The approach for method selection presented in the following is centred around the identified failure mechanisms, see {numref}`meca-table4_6` giving an overview of different mechanical part types with associated failure mechanisms and modelling recommendations. The third column of {numref}`meca-table4_6` gives an indication about the relevance of each part type for different classes of mechanical systems introduced in {numref}`meca-table4-3`. As has been discussed in {numref}`meca_9_4`, it can be sufficient to focus on those parts that dominate system level reliability. In the same way, the reliability of a specific part is often dominated by a single (or very few) failure mechanism(s), allowing to focus the analysis especially with methods that are based on the PoF. Which failure mechanism(s) dominates reliability depends not only on the part type, but also on the application, e.g. a spring is likely to fail differently when used with static, cyclic or dynamic loading. The “predominant” failure mechanisms listed in {numref}`meca-table4_6` should therefore be considered only as a first guidance. The remaining columns of {numref}`meca-table4_6` provide recommendations on the modelling of the selected parts and failure mechanisms. The focus is on the methods making reference to the PoF, in particular Structural reliability methods as outlined in {numref}`meca_9_6` and {numref}`meca_9_7`. The application of the handbook approaches is restricted to selected parts, allowing the possible use of the {term}`NSWC` handbook {cite:p}`mec-NSWC-11` as a scaling method (which is based the {term}`physics of failure `) to adapt statistical estimates based on data samples that are more relevant for space applications than the {term}`NSWC` data, see {numref}`meca_9_3_1_5` for discussion. The {term}`NPRD` Handbook {cite:p}`mec-NPRD-16` is not considered any further. In {numref}`meca-table4_6`, statistical methods are mentioned only where a prediction based on a pure PoF approach is unlikely to be successful. However, as has been discussed already in {numref}`meca_9_3_2`, , statistical methods are generally preferred over PoF approaches provided that sufficient and relevant data is available to perform the prediction. The failure mechanisms listed in {numref}`meca-table4_6` can provide guidance for judging the relevance of a specific data set, always keeping in mind the strong variability in mechanical parts reliability when seemingly identical parts are applied under different conditions. Also, a combined approach with Bayesian updating of a prior derived from the PoF, using relevant data even with limited sample size, is generally preferred over a pure PoF approach, see {numref}`meca_9_3_2_4` and {numref}`meca_9_6_5` for guidance. ```{list-table} Mechanical part types, associated failure mechanisms and modelling recommendations :name: meca-table4_6 * -

					Structural reliability methods			NSWC 2011 handbook
Part type	Description	Mechanical item category	Predominant failure mechanism	Modelling recommendation	Potential failure mechanisms	Limit state function	Simplified method	Model available	Space relevance of NSWC method
Bearings	A device that constrains relative motion to only the I Active desired motion, and reduces friction between moving parts	I Active	Lubrication wear	Structural reliability methods	Wear Fatigue &fracture Distortion	generic			Method not applicable due to predominance of lubrication wear over fatigue in space
Springs	An elastic object that stores mechanical energy, II Passive used to exert forces or to provide flexibility.	II Passive	Load loss, creep (distortion)	Statistical methods, NSWC adjusted, or Structural (but difficult to model creep)	Fatigue &fracture Distortion				Depends on the type and application; fatigue/fracture less relevant in space
Gears and splines	A toothed machine part such as a wheel or cylinder I Active that meshes with another toothed part to transmit motion or to change speed or direction	I Active	Lubrication wear	Structural reliability methods	Wear Fatigue &fracture Distortion	generic			Application questionable due to space specific lubrication and rotation speeds
Shafts	A rotating device, usually of circular cross section that transmits torque from one part to another	I Active	Wear (when contact is present),otherwise fatigue	Structural reliability methods	Wear Fatigue &fracture Distortion	generic			Method is maintenance related and thus not applicable to space
Hinges	Ajoint that holds two parts together such that one II Passive can swing relative to the other	II Passive	Increased friction or kinematic issues	Structural reliability methods	Distortion Wear	generic			Only some general considerations in NSWC
Mirrors and lenses	A piece of glass or similar material designed to reflect (mirror), or to concentrate or disperse light rays.	III Structural	Fracture	Structural reliability methods	Fatigue &fracture Distortion Degradation	generic			NA
Flexural pivots	A flexure which is engineered to be compliant in one or more angular degrees of freedom.	I Active II Passive	Fatigue	Structural reliability methods	Fatigue &fracture				NA
Fusible Wires	A device consisting of metal strips soldered with a fusible alloy that melts at a specific temperature, allowing the pieces to separate	II Passive	Quality issue or random failure	Statistical methods	Fatigue &fracture Distortion	generic			NA
Seals and Gaskets	A device placed between two surfaces to prevent the flow of gas or fluid from one region to another	I Active II Passive IV Fluidic	Wear, degradation, installation damage	NSWC adjusted, Statistical methods or Structural reliability methods	Wear Distortion Degradation	generic			Method may be applicable but requires base failure rate adjustment
Structural component (ind.threaded fasteners,inserts, brackets)	A mechanical component designed to sustain loads or pressures, provide stiffness or stability, or provide support or containment.	III Structural	Fracture	Structural reliability methods	Fatigue &fracture Distortion	generic			NA

``` (meca_9_5_3)= ## Selection of methods for assembly level mechanical reliability prediction Reliability prediction for assemblies of mechanical parts can either be made bottom-up, combining part level inputs with system level reliability methods, or top-down, using inputs directly collected at the level of the considered assembly. In general, similar considerations apply as for part level mechanical reliability prediction, leading to the following modelling recommendations: * **Statistical (or combined) methods** are suitable provided that relevant failure data is available directly at the level of the considered assembly. * **Physics of failure methods** based on Structural reliability theory are suitable if the assembly level probability of failure is dominated by one or very few {term}`failure mechanisms ` (see also the discussion in {numref}`meca_9_4`). * **Bottom-up modelling** starting from part level analysis is recommended in case none of the abovementioned conditions is fulfilled. ```{admonition} Note :class: note The NSWC Handbook {cite:p}`mec-NSWC-11` contains models also for some items that are better considered as assemblies rather than parts, including e.g. electric motors, solenoids and transducers. For reasons already discussed in {numref}`meca_9_3_1_5`, the direct use of these models is not recommended for space applications.