Bending moments transfer through adhesive bonded joint 1

[adampar]

Hi respected engineers!
I face a design issue in which i would like assistance due to my lack of experience. I want to connect a CFRP tube with metal inserts at each edge (the metal fittings are of smaller diameter than the CFRP tube and length of 33mm into the tube) and the proposed joint is an adhesive bonded one. After searching on the net, I found that these joints are mostly used to transfer axial-torsional loads. Some preliminary calculations done, have shown that the peel and shear stresses are too high to sustain a bending load, as well as the out-of-plane strength of the CFRP tube is also low. Just to provide more info, the tube will act as a cantilver beam of approximately 600mm, and a max moment of 60Nm must be tolerated. What type of joint would you advise me to use to transfer this moment? Could this be done with some alternation of the design and still use adhesive a the connnection or is it mandatory to use a mechanical type connection?

------This is a repost from Structural Engineering Forum (

http://www.eng-tips.com/viewthread.cfm?qid=408172),

where SWComposites suggested me to post here------

 

[adampar]

Also to reply to SWComposites and racookpe1978 suggestions from the old post, the tube is sized according to the bending loads, its flexural properties are good enough to carry that bending moment, but when it comes to interaction with an insert, high out-of-plane shear stresses occur, which can be devastating for the composite material. I agree that tests should guide me towards the solution, but i would like to know beforehand whether it is viable to use metal inserts. I am also thinking maybe a filament winding overlap of the bonding area, or similar techniques woould improve the situation.

 

[adfergusson]

A well designed adhesive bond should pretty much always be able to out perform a mechanical joint. However, the design of adhesive joints is rarely the straight forward matter that many people seem to think it is. To get an idea of an analysis process for a highly idealised (I.e. Pure mode 1) adhesive bond, have a look here: Link.

There are also some other resource on that website that are worth a look and I've scribbled some notes and diagrams (see attachment) that should give you some food for thought too.

 

[blakmax]

Had a look at the link adfergusson provided, and I immediately dismissed the work. If anyone believes that they can assess the performance of adhesive bonds which use grit-blast and solvent degrease as the complete surface preparation, they are sadly mistaken, I don't care how great the mathematics is, if you are measuring garbage, your results will be garbage, no matter how you measure it.

The most significant mechanism that drives interfacial degradation is hydration of the surface oxides. After grit blasting, the surface will rapidly oxidise and when the adhesive is cured chemical reactions occur between the adhesive and those oxides. If the oxides are prone to hydration (e.g. Al2O3 forms Al2O3.2H2O) then the chemical bonds formed when the adhesive is cured dissociate to enable hydration to occur. The rate of dissociation depends on the presence of water molecules and the time since the bond was formed. Hence, the result obtained from ANY mechanical test will depend totally upon the extent of hydration that has occurred at the time the specimen is tested. This is true for any strength, fracture or fatigue test. So the result will depend more upon the time the test was undertaken, and any conclusions drawn are unsafe.

I have never had confidence in fracture mechanics for bonded joints because there are two basic failure modes; cohesion where the adhesive fractures and adhesion where the failure propagates along the interface. In between these extremes, there is mixed-mode failure which is a mixture of adhesion and cohesion failure. (See

http://www.adhesionassociates.com/p...res - Mixed-Mode Bond Failures Explained.pdf)

It is impossible to define a single fracture parameter that describes the different failure modes, let alone the mixture of failures that occur in real bond failures and then to realistically add TIME SINCE MANUFACTURE as a design parameter.

It is far more productive to ensure that the bonding process prevents hydration, and I can assure readers that grit-blast and solvent clean can never prevent hydration on most metals.

I suggest before you start your project, read

http://www.tc.faa.gov/its/worldpac/techrpt/artn0657.pdf

.

Regards

Blakmax

 

[adfergusson]

BlakMax; I initially thought you had misread, or simply missed, large parts of the content in the links. However, having looked at the papers you linked to it then became apparent that whilst you might have had some insight and experience that would have been uncommon back in the 80s, you appear to have buried your head in the sand for a couple of decades only to re-emerge now whilst failing to grasp that, amongst other things, the fields of adhesion, fracture mechanics, adhesives, computing, numerical methods, etc… have changed dramatically over the decades that have past. Most obviously, you appear to think that knowledge of the need for surface treatment of substrates prior to adhesive bonding is some sort of expertise when in today’s day and age it is common knowledge within the relevant fields; a whole range of methods for many substrates, environments, etc.. are openly described and available on the internet and these are also matters your adhesive and/or substrate supplier will likely be able to advise on (for free) through their technical sales support or on-line documentation and application guidance resources, e.g; LinkLink

‘Had a look at the link adfergusson provided, and I immediately dismissed the work’

Well the you’d better avoid getting in any modern car as the approach described by Ambrose (Disclosure; Ambrose and I co-supervise some work together at Imperial College) in the first is the basis for the design methodology for structural epoxy adhesives used in most cars. These design techniques were developed by members of the adhesives group at Imperial College, including Tony Kinloch and Gordon Williams (both Fellows of the Royal Society with Tony holding the position of Chair of Adhesion) and first used in Lotus’ bonded chassis.

‘ If anyone believes that they can assess the performance of adhesive bonds which use grit-blast and solvent degrease as the complete surface preparation they are sadly mistaken,’

You seem to imply that I, or the document in the link, am suggesting that this alone is acceptable? If you actually looked at Ambrose’s document (which is now twenty years old) then you’d have seen it that the work involves comparing the fatigue life of joints with different surface treatments tested in dry and immersed in water (with comments on water diffusion rates).

I don't care how great the mathematics is

That documents sticks to mode 1 testing as the mechanics behind this load state is comparatively simple; the maths contained therein are limited to first order differentiation and integration. If you want to do an Engineering degree in the UK then you need to be able to solve these sorts of equations whilst you are still in high school. You propose on your adhesion associates website that you have some special blend of analytical solutions that allow you to accurately predict joint performance; given that all but the most very basic load states in components made of only one material can be reduced to one dimensional problems, or ‘great mathematics’ as you might call it, how the hell do you go about dealing with real world structures where the stress states require description through higher order partial differential equations?! Or, in addition to fracture mechanics, do you not believe in Hooke’s law and the general framework of elasticity either?

Hence, the result obtained from ANY mechanical test will depend totally upon the extent of hydration that has occurred at the time the specimen is tested.

Yeah, sure, the mechanical and thermal properties of the adhesive and substrate have no bearing…. nonsense…... just utter nonsense. Unless that is (as blakmax appears to) you come from a time and place where the idea of doing (now antiquated) surface treatments such as a chromic acid etch or anodization would be thought some unusual novelty.

cohesion where the adhesive fractures and adhesion where the failure propagates along the interface. In between these extremes, there is mixed-mode failure

I noticed that you cited one of Tony Kinloch’s book (Kinlock, A.J., Adhesion and Adhesives, Chapman and Hall ltd, New York, 1987, pp. 78.); was page 78 the only page you read? Other parts of the book discuss mixed mode behaviour. Mixed mode behaviour in the context of the adhesives (and composites) community is understood by pretty much everyone, but you, to mean the proportion of the stress state at a crack that is mode 1, 2 or 3. You appear to have some bizarre little convention of your own but, for example, the book of Tony’s that you cite has been cited 2181 times (https://scholar.google.co.uk/citations?user=bXUqSnEAAAAJ&hl=en) and the nature of mixed mode failure is described in there as the ratio of stress components at a crack tip. Additionally, I’ve never come across someone describe/misunderstand mixed mode fracture behaviour in the way you have; perhaps this is why the only person who seems to cite your publications is yourself? If I am wrong then please point me to a peer reviewed journal paper where this description of mixed mode behaviour has been accepted.

It is far more productive to ensure that the bonding process prevents hydration, and I can assure readers that grit-blast and solvent clean can never prevent hydration on most metals.

This is such a well-known ( I was introduced to the subject in the second year of my UG degree) issue these days and loads of companies have standard processes off the shelf solutions for most scenarios . Adampar; I suggest you make contact with your local adhesive supplier and/or speak to small scale MROs around your local airfield/airport in your area who can advise you on companies that can carry out etching or anodising suitable to your substrates. Now-a-days there are also alternatives to these unpleasant processes that rely on relatively benign chemistry, like functionalised silane and siloxane chemistry that are also highly resistant to moisutre and passivate substrate surfaces from hydration. Blakmax would probably be horrified by these too (even if the likes of Airbus, Boeing, Rolls- Royce, etc.. use surface treatments like these) as some of them are water based colloids

 

[blakmax]

Wow...Don't hold back, mate. Tell us what you really think!!!

I choose to ignore the personal diatribe and as just a few simple technical question:

1. Any computational method is only as good as the fundamental assumptions on which it is based. So how do you apply LINEAR ELASTIC fracture mechanics to a material which derives between 50 to 80% of it's strain energy to failure from plastic behaviour? (Look up any of the myriad of Hart-Smith's papers on bonded joints). The same criticism can be leveled at elastic Finite Element Analysis of bonded joints.

2. As I stated, what is the sense of undertaking fatigue tests to evaluate bad surface preparation methods when the result will change if the specimen is allowed to sit around for a few months while the interface hydrates? Testing in dry/wet environments is meaningless because short term moisture conditioning of specimens only has an effect on the bulk adhesive properties, and has minimal effect on the hydration of the interface. Interfacial degradation depends totally on time since manufacture. Try it some time. Take a badly prepared specimen batch. Test some soon after manufacture in dry or wet environments. Let the rest sit out in the rain/sun/snow for twelve months and test again. If the results are not dramatically different (wet or dry) then I will happily buy you a dozen bottles of your favourite beer!

3. I have never, ever advocated chromic acid etching, not would I. How you extracted that from my comments or my papers is bewildering.

4. My "whacked out theory" of hydration and mixed-mode failure is supported by research. see Nitowski, G.A., Topographic and Surface Chemical Aspects of the Adhesion of Structural Epoxy Resins to Phosphorus Oxo Acid Treated Aluminum Adherends, PhD Dissertation submitted to the Faculty of the Virginia Polytechnic Institute and State University, August 26, 1998, Blacksburg, Virginia

5. Everyone knows that it is exceptionally bad engineering practice to design bonded joints to be loaded in Mode 1.

6. I am well aware of the use of such materials as organofunctional coupling agents for adhesive bonding and indeed I was part of a team that pioneered the use of these materials for on-aircraft bonded repairs. In fact, applying the outcomes of our work to real, practical repairs, I was able to reduce the repair failure rate over a period of fifteen years from 43% to almost zero, and in the cases where we did experience failures we were able to determine where the technician had taken a short-cut. SO in all cases where the correct methods were applied we had ZERO bond failures. The USAF reports similar results.

I hope you can address these issues in a more rational impersonal professional manner. If not, go back to the trolls on Facebook!

Regards withheld

Blakmax