CX in load transfer joint 2

[Fatstress]

I am almost convinced that no fatigue life enhancement can be obtained when CX (hole coldworking)is applied on the fastener holes of a medium or high load transfer joint. A significant fatigue life improvement can be achieved when it is applied on an open hole where the mode of failure is clear (fatigue at the root notch due to repeated stress peaking). But not for the case of riveted/fastened joint which can be failing due to other mode of failure such as fretting. Anybody has the same opinion ?

 

[rb1957]

i'm sure the folks at FTI who market CX solutions would disagree.

i would also say (a little like you comments) that in practice CX may not help a joint that fatigues due to fretting, but from an analysis prespective comparing a CX hole with a plain hole is not worth it (the CX is going to be far superior).

in my experience i wouldn't CX every hole to get the benefit, I'd prefer to get the design 95% right and use CX in a targeted manner to deal with the remaining 5%.

 

[Fatstress]

I agree with you rb1957. Better get your design right first, and apply any mechanical process improvement as an additional assurance only, for example when your analysis result is marginal. But the question is, can such mechanical processes even degrade the performance? Or is there a restriction or recommendation where this can be applied? When you perform a hole coldworking, I would imagine there will be a slight local material bulging at the mating surface which may cause a problem under cyclic loading.

 

[FastMouse]

Hi there. I have been involved in test programs that looked at the effect of cold expansion on fatigue life. These were coupon test programs, and we had a configuration that we considered to be representative of high load transfer joints in our aircraft. We saw definite improvements in fatigue life due to cold expansion when compared with non-cold expanded HLTJ holes. The alloys were 2000 and 7000 series and bolt diameters were in the approximate range of 0.25 inch to 1 inch diameter. Fastener installation was both clearance fit and interference fit. The level of cold expansion applied was a function of hole diameter, decreasing for the larger diameters.

The improvement in life was dependant on several criteria, including alloy type, hole diameter, stress level, and stress ratio. We had significant confidence in the fatigue benefits of cold expansion and used the process to increase allowable stress levels, and thus reduce material thickness and weight. Of course, this did reduce the scope for in-service reparability, or modifications to deal with unexpected fatigue issues, but such is the way of things; heavy and conservative or light and less so. The reduced scope for remedial action in the event of in-service issues meant that it was very important to really understand the expected fatigue performance of the structure in the design phase.

Fatstress is correct, there was local bulging of the material local to the periphery of the hole. This would exacerbate fretting, but, nonetheless, the empirical results showed improvements in fatigue life. However, a question to which there was never any definitive answer involved aging: since the residual stresses resulting from the cold expansion are so high, are they reduced by material creep over time, say a few years, resulting in a reduced the fatigue benefit?

FastMouse

 

[WKTaylor]

Fatstress, RB1957 and Fastmouse...

In my opinion, You raise some interesting points that are not well defined or annunciated by current engineering knowledge and practices.

I would like to know Your [collective] simple definitions of what low, mediun amd high load transfer joints are. I have my definition, but I'm curious as to >Yours<!

Question: what are Your experiences with the CX process I am most familiar with (due to recurring use) as follows:

Conventional CX: hole cold-expansion "X"%; then finish-ream to fastener hole size.

CX2S: hole cold-expansion to precision diameter, install fastener without reaming.

ForceMate: Cold expand a bushing in a precision-reamed hole, the finsih ream the beaing ID to size.

Also... The natural aging affect of temperature cycles on aluminum is worriesome to me.

In the early 1970's my dad completed a T-18 homebuilt airplane [all metal]. He set aside some 2024-T3 sheet in a garage with direct sunlight exposure in a high-desert environment. About 15-years ago I grabbed a piece of 0.016 thick sheet intending to make a simple bent bracket: It cracked when I bent it over a 0.09R edge and "felt" vey stiff [high yield]. I played with several pieces... then realized that it was probably cloer to -T62 than -T3 in temper. Subsequent literature indicates this "aging" happens to sheet metal in high temp climates.

On the military acft I now work on, we have a lot of 2024-T3, 7075-T6 and 7178-T6 sheet [service temperature limit of 300F and 200F x2 respectively]. I have a test-report that was written in support of electronics heat problems in a desert climate. A lusterless-dark gray aircraft was thermally evaluated thru mid-late-day at the NTC [Mojave dester] in CA: body skin temps exceeded 175F and were rising, when the exterior IR thermography testing was halted.

So the question arises in my mind: has the aluminum overaged from -T3 to -T6 or -T6 to -T73... or worse? I can't get a clear answer from any engineering group, because no study of current hardness/conductivity exists... and no-one seems concerned. this raises the next issue: what is the effect of long-term desert heat exposure on CX'ed 7075-T6 or 7178-T6? As for our acft: the normal hole-fastener fit is "transition [net-fit], due to suseptibility of the 7xxx-T6 alloys to SCC. The CX2S [and obviously ForceMated] holes see a definite sustained interference: does this CX indeed maintain itself... or creep-relax... over time?

Comments? Experience?

Regards, Wil Taylor

 

[Fatstress]

Fastmouse and wktaylor thanks for your reponses,

Fastmouse, thanks for sharing your opinion and experience about this subject. In fact, I also have been involved in the evaluation of test results of hole coldworking on lap joint specimens. But I am afraid I have different result than yours. Well, for 7000 you can have some improvement but in our case not for 2000. Different fastener types, interference, specimen thickness were also investigated. I wonder why do we have different results. Have you checked the failure modes of your failed test specimens ? Are they failing starting at the net section of the hole (9 and 3 o'clock) or a bit moved upward (10 and 2 o'clock). Or maybe the crack initiated a few mm away from the hole ? Aging effect on residual stresses ? I would think it is relevant. But I guess you need to have a lot of patient to carry-out a test to see what the effect would be.

Wktaylor, low-medium-high load transfer joint. I think I know it when I see it. I would say, a (fastened) lap joint is a high load transfer joint and a stringer and maybe a shear clip to (fuselage) skin joints are low load transfer joints. Any joint configuration in between, is a medium load transfer joint. In term of the amount of load being transferred, 25% to 50% load transfer would be a high load transfer joint, 5% to 25% medium, and less than 5% low. In general a high load transfer joint would be a joint where a significant amount of load is transfered as bearing load. A significant difference of mode of failure compared to an open hole should be expected.

A CX process for me is to prepare the hole to be coldwork, install the required sleeve, slide-in the mandrel, pull it out, and ream it to the final diameter (I wish we can use the sleeve again).

Aging effect on residual stresses obtained by coldworking and the tempering condition of a material. My engineering judgement would say, yes there will be an effect. But I think we have to invite material experts into this discussion. I am sure they have the answer.

FATSTRESS

 

[FastMouse]

Fatstress: Failure modes in our specimens were not unusual in the vast majority of cases. Cracks initiated at the hole periphery, perpendicular, or almost perpendicular, to the applied load direction. Failure sections were subjected to a quick fractographic inspection to confirm that the failure was typical. We had good results in both 2xxx and 7xxx alloys. Perhaps there is something about the material thicknesses used, or differences in fastener installation standards, etc, the lead to different results, or maybe it was to do with loading. So many variables…

wktaylor: We used to categorise joints as either HLTJ or LLTJ. We didn't talk about MLTJs. Our HLTJ fatigue coupons were arranged so that all the load in the joint was transferred by two equally-loaded bolts, so I guess that our definition of a HLTJ was one in which 50% or more of the load was being transferred through a single fastener.

Our LLTJ specimens also contained two equally-loaded fasteners, but not all of the load in the joint was transferred through the bolts, so the total bolt load was less than the load in the joint, and thus the load transfer per bolt was less than 50% of the load in the joint. Regrettably, I do not have ready access to the percentage load transfer that this configuration gave us. Maybe other organizations would have classified these as MLTJs.

For conservatism, we used to analyse skin-spar joints using HLTJ data, even though the load transfer per fastener was less than 50% of the joint load. Skin-stringer attachments were usually analysed using LLTJ data. Fittings were usually treated as HLTJs.

Maybe the ratio of bearing stress to net stress is a better way to charaterise what is happening in the material, rather than percentages of load transferred.

As far as Cx processes go, we usually used 4% Cx, less for large holes, and always performed a final ream operation. There was some minor discussion about Cx to final size, but that had not resulted in any conclusions by the time that my fatigue work ended at that organisation. Additionally, although we were aware of the ForceMate system, we did not take into account any fatigue benefit in the calculations because we had no auditable data for it. We used it only in selected locations, and not for fatigue reasons.

The comments about temperature cycle effects are interesting, although I regret that I can not add anything meaningful to that discussion. All our testing was performed in ambient lab environments, and the subsequent, statistically-processed results were used directly in our analyses. I am not aware of any artificial aging of the material prior to testing, although perhaps there was some; I just don't know. Furthermore, we had temperature probes on the coupons to ensure that they did not get too warm during the rapid constant amplitude load cycling.

Another interesting temperature question concerns the effect on fracture toughness. It is common knowledge that reduced temperature tends to result in reduced fracture toughness. If you have a limit load case that occurs at cruise altitude (I'm thinking commercial transports here) when the structure will be cold, how can you ensure that a crack of a given size will not be critical when the you have only room-temperature fracture toughness data to work with? One answer that was suggested to me was that the issue was not a great concern because, in practice, actual, real-life limit load encounters tended to occur at low altitudes!

 

[rb1957]

regarding the temperature effect on fracture toughness ... i don't think it has a significent (on commercial transports), 'cause the key result is the repeat inspection interval which won't change much due to a decrease in toughness ('cause the crack growth rate is so rapid as you get close to critical). it (low temperatures) might be more significant in considering the reduction on crack-arrest capability.

 

[Fatstress]

Fastmouse,

I am still curious of your test results. Your high load transfer joint type of specimen used in your test, was it a lap joint, a single shear butt joint, or double shear butt joint ? Or do you have a significant secondary bending moment in your test specimens ? If it is a symmetric type of joint, then I am in line with your findings. The secondary bending can really make life difficult.

rb1957,

fracture toughness is usually correlated with crack growth rate of the material, because both somehow represent the ductility of the material. If the fracture toughness is decrease I suspect that the da/dN-dK will also be degraded, therefore the crack growth life, and finally the interval inspection will be reduced.


Fatstress

 

[rb1957]

good point ... sent me checking AR-MMPDs-01 (or MIL-HDBK 5) ... Al alloys (ok, I looked at 7075 'cause i had it open) increase fty with temp < r.t. ! ... steels reduce fty as they go brittle at lower temps

 

[FastMouse]

Fatstress,

The HLTJ coupons were double shear lap joints. Hopefully I can make a sketch with text characters…


FF FF
-------------------------------------------------------------
P/2 <====== |/////////////////////////FF////////////////FF////////////|
-----------------------------------------------------------------------------------
|//////////| |//////FF////////////////FF/////////////////////////////////| ============> P
|//////////| |//////FF////////////////FF/////////////////////////////////|
-----------------------------------------------------------------------------------
P/2 <====== |/////////////////////////FF////////////////FF////////////|
--------------------------------------------------------------
FF FF


FF indicates fastener position. The sketch is nowhere near scale! The total thickness of the two outer ligaments was a little less that the inner layer, so they had slightly higher stress, but so no much so that the failure was always in the outer ligaments. The joint is symmetrical, with no secondary bending.

The point about crack growth rate as a function of temperature is a good one. On the other hand, if you have a structure where the crack growth is driven primarily by the GAG, it might be possible to argue that room temperature da/dN data is applicable. The structure is initially loaded when it is warm at the start of the mission, is subjected to the less important residual gust part of the spectrum when it is at altitude and cold, and finally unloaded when it is warm again at landing - so the majority of the loading that causes crack propagation occurs during warm mission segments. An example of what I mean might be wing skin/spar joint in a dry bay, which is loaded to, say, 85% of the GAG max stress on rotation, and then unloaded at landing, both "warm" mission segments. The residual gusts and the final 15% of the GAG max stress occur when the structure is cold. On that basis, there is perhaps scope to argue that room temperature da/dN data is broadly applicable. Wing-mounted engine attachments, which are sometimes primarily loaded in fatigue by gust cases are maybe an example where that reasoning fails if those gusts are occurring at altitude and the structure is cold.

The more I think about this, the more it is possible to extract really interesting points from the details. There has got to be a PhD in it for someone.

 

[Fatstress]

Fastmouse,

That's it, our specimens are different.
I guess I was not specific enough regarding my suggestion that no life improvement of hole coldworking for high load transfer joint. It should be meant for joints with significant secondary bending moment.

Your point regarding the warm GAG cycle is quite interesting.
Sometimes ago I was curious whether the rotation was causing such a big bending moment to the wing structure as what was suggested by the loads people. Then we decided to monitor our Strain Gages installed in the flight test aircraft wing structure from standing on the ground, engine run-up, taxiing, take-off run, rotation, climb, cruise, and...we concluded the bending load prediction for rotation is too conservative. I would say at rotation the BM should be not more than the maximum 1G BM of cruising or descent.
Nevertheless, I think there was a test already carried out showing that for common Alu alloy no significant change on the crack growth rate occurred untill -50 C. But of course it was an accelarated test of 5 to 10 Hz.