Material Variability Factor for Composites

[irq]

Hi,

After a long absence on this forum I want to continue this thread16-154238

There was a disscussion about taking into account a "material correction factor" to the ultimate test load value according to 25.307(d).
This paragraph implies, that no additional correction factor must be included in case if the tested part "has features such that a number of elements contribute to the total strength of the structure and the failure of one element results in the redistribution of the load through alternate load paths".

Consider there is a joint, where the metal fitting is connected to the composite structure by means of 6 fasteners, which were identified as a weakest members of the connection. The fastener-to-composite structure was tested and the B-Value from a coupon tests estimated obtainig a design allowables on which the joint design and final amount of fastener connections is based. At the end one sample of the joint (fitting installed to the composite plate with 6 fasteners) was tested (limit*1.5) to verify the joint analysis, where a failure load exceeded the ultimate load only by 1-2% (!).

According § 25.625 (a) and 25.307(d) no additional "fitting factor" or "material correction factor" need to be applied, as the test was performed in actual relatively simply and well known stress conditions and there is a load redistribution via 6 fasteners. Due to a brittle nature of composites and no ductility effects I have thought about the latter point to answer the question if we pass the test. What do you think does this point apply to the composites and can we avoid this reduction factor or there are other aspects, which should be considered?

Do you apply the A-/B-Value concept in term to check the variability of the results to the element or joint testing or on the material properties level (coupon tests) only?

[irq]

 

[SWComposites]

So is the joint test critical in the fasteners, the composite plate or the metal fitting - the prediction and the actual test result? Is the joint test fully representative of the actual part structure, or a generic joint?

 

[irq]

The joint is critical in proximity of the metalic fasteners and the connection fails obviously in composite plate, which confirms our prediction from a coupon tests (see above post). The joint is representative, original fitting & fasteners as installed in the component, identical production process etc. The stiffness arround the joint is also similar, the load distribution on the fastners deviate by let's say no more than 5% from the actual part structure (identical is not possible becouse only a plate and not the real structure behind the joint was simulated). Even when the fasteners are mostly loaded in shear, the failure severity is not identical on all locations, which arised the question about the load redistribution and applicability of the factor 1.0 as noted in 25.307(d).

 

[SuperStress]

Irek,

I have not seen A- or B-basis calculations applied to joint allowables in the way you describe, only to material property or fastener strengths. For single load path structure, A-basis numbers should be used. B-Basis should only be used for redundant structure or repairs/doublers, etc. I am not intimately familiar with the generation of composites allowables, I'm primarily a metallic guy.

The sentence in 25.307(d) is essentially saying that there has to be a redundant load path if this one fails; that is, that the joint is fail safe. If this joint fails, can load safely be redistributed to other structural members without catastrophic failure?

The 1.15 fitting factor mentioned in 25.625 should be used in any joint not proven by limit or ultimate load tests, assuming that production processes, tooling, etc. were used and that all loading and boundary conditions are representative of the joint. The part that concerns me about your post above is the fact that you say a plate was used to simulate the backup structure, which did not exactly match the stiffness of the production configuration. Unless you can confidently quantify how much your stiffness deviated from the production configuration, and how that stiffness change would have affected the joint load distribution, I would think that the 1.15 factor per 25.625 would still apply.

Let me know if that answered your question.

SuperStress

 

[SWComposites]

Do you apply the A-/B-Value concept in term to check the variability of the results to the element or joint testing or on the material properties level (coupon tests) only?

> if the joint test is to certify the specific joint that is being tested, then generally no variability factor is applied, assuming that you also analyze the joint using A- or B-basis allowables and show +ve margins at DUL.

> if the joint test is to be used to develop some sort of allowable or design value or factor to apply to an analysis of other joints, you may have to apply a variability factor.

> it all depends in the details of the specific situation, the fidelity / quality of your stress analysis, the amount of allowables test data you have, your experience in predicting previous subcomponent test results, and who the cert agency or customer is.