Keels and Keels Again!

An intersting article on keels and rudders engineering, by our fellow member Eric Sponberg:
http://www.proboat-digital.com/proboat/200508/ (Search Content)

I agree with Mike John on that keel root attachement not seeming to be a properly engineered one. Also those bolts seem to have been overtightened.

Down here a couple of images of what in my opinion would have been a better engineered keel root for a high aspect bulb keel. Another way would have been to fit the top of the keel into a socket built into the boat and bolt the keel horizontally through the socket, as Eric suggests.

Cheers.

 

I spent the day sailing on a boat that is about 15 years old and has a smaller footprint for the weight/lever arm than the boat we have been discussing. Neither that boat, nor any of her sisterships, have ever had a keel issue. These boats have been sailed hard over the years in up to 30+ knots and big seas.

IIRC, the keelson was 5/8" on the cynthia wood. As reported on SA or the Houston Chron. Perhaps, Paul, the boats you are familar with are considerably thicker (stiffer) in that region.
I am only guessing here, perhaps the prior grounding was a lot more serious than was detected.Or the laminate poorly constructed. ( resin rich) Although a professional appraisal and repair was supposed to have been commissioned.
I have no problem with this design as a weight saving procedure as long as it is accepted for what it is and the ramifications if the keelson laminate can allow an instant detachment. No pun intended.

I don't see any of your problems with continuous timber floors, heavily laminated to the hull with the keel bolts in epoxy anti-compression annuli. Not a light solution tho'.
I have followed David Gerrs scantlings on my build and they are nothing like the CW.

Based on some opinions that should not be possible. It only blew in the low 20s today, so maybe we were lucky.

I don't beleive anybody actually said that it should not be possible.
What was implied was, a catastrophic failure with little or no warning could occur by nature of the potential failure.
I think the issue is, that with better engineering,a lot more time is available before the boat threatens the crew.

 

Too clean the cut and no traces of the supporting structure. Something was very, very wrong.

Cheers.

 

The "guesstimate" of 5/8" was made by eye, with no measurements done, by someone who may or may not have the credentials to give an accurate estimate.

Plane off a section of timber to 5/8" and walk around asking a representative sample of people to tell you the thickness. What do you think the range of the sampling might be?

I have my own questions regarding the design, build, and repair of the boat in question, but I'm sure not going to start hurling unfounded remarks as some people here have. I can understand seeing it on some other message boards, but this is supposed to be a group of people interested in design.

There will be ample opportunity to burn the witches once the facts come out. However, the genie can never be put back in the bottle if unfounded criticism is published against a designer who can't defend himself in public until after things are sorted.

Let's for a moment assume the laminate is only 5/8" thick. For all the loud talking that's gone on I have yet to see anyone attempt to show that the laminate thickness was inadequate.

What sort of boat have you built? I'm not familiar with Gerr's method, or of any boats he has designed. Can you fill us in on how it is done?

As far as the anti-compression tubes, what keeps them from pushing their way through the hull laminate?

My own boat is cold moulded in the West System epoxy method. My keel is a higher aspect ratio than the boat in question. The keel bolts originally came up through the wood keelson. That keelson was later capped with an aluminum H beam. Having longer length keel bolts so you can go up through thick material buildups can lead to problems in itself. The shorter the bolt the better, of course, unless the entire structure is incredibly stable.

Oh, there have been comments about this construction method being totally inadequate, but little in the way of facts. Of the 10 or so sisterships to this boat, have any had any sort of keel issues?

One other thing to think about, regarding the idea to use a metal frame in every boat with an externally bolted fin keel, is how that would work if you wanted to use a carbon fibre laminate. I think I can hear the fizzing now...

 

Please tell us how you conclude the bolts were overtightened from that photo.

A keel flange is not a bad idea for this type of keel, but you have not attempted to address the failure, which is in the supporting structure.

You have also not mentioned how this will not work on a boat with a sump (the majority).

I was going to give partial credit for this attempt, but then I saw this:

Negative points there, so total score: nil.

 

So why have you not enlightened us?

So much bluster on this board about how poor the design is, yet not one real alternative from most of you.

 

One thing at a time, my darling.

A rule of thumb says the laminate in way of the keel should be: emin= 2.8*Lwl/1.8 mm

Lwl is 10,36 m for the Cape Fear 38, so emin = 17,38 mm, let's make it 18 mm.

We can asume the eye estimated 5/8" (+/- 16 mm) could in fact be 18 mm, so the guy was about right.

We are dancing, Paulie Bee.
Love.
Willie Boy.

 

There is some interesting discussion on the design of the Cynthia Woods hull-to-keel connection in the last few pages.

I have experience with similar connections in an industrial application with steel to steel parts and have personal experience with a cast steel keel to fibreglass hull failure that was a combination of my abuse and lack of analysis on the designer's part.

With the industrial application my first impression was that the connection was poor because there would be edge loading along the line of contact between the top edge of the butting plate to the flat face. In fact this only occurs if there is insufficient preload in the bolts. The problem was remedied simply by fitting high tensile bolts of the same size and accurately torquing them up on installation. The analysis was not difficult as I paid for FEM of the joint to see what was going on and both parts were steel plate being almost isotropic.

The situation with the keel is complicated by the different materials. Also the elastic properties of glass composite vary with the direction of layup. In the plane perpendicular to the layup the elastic modulus is about 1/4 of the modulus in the direction of the layup for a mat material. All three axes vary if it is a directional layup.

The elastic modulus of lead is about the same in the direction of layup so around 4 times higher than glass composite in the perpendicular. My thought on the consequence of this is that, as the cantilever load is applied, the lead will concentrate stress in the glass toward the edge disproportionally as the lead is not as compliant as the glass. Hence there is a guillotine type action.

The situation where the keel is mounted off an internal or external step in the layup is somewhat different because the stress is aligned more in the direction of the layup hence the properties of the glass are closer to the properties of the lead. This reduces the stress concentration along the edge but also there is more give in the structure to further reduce the stress concentration.

If the keel has a higher elastic modulus than the glass and the keel is highly stressed (as in thin) then there needs to be a wider flange at the connection to reduce the stress in the glass. This is common practice with steel keels I have seen.

Modern engineering practice for a joint as critical as keel to hull would be based on knowledgeable FEM of the actual connection.

SO... Who can do a model of the failed connection and compare with the stepped case? I expect even a small step would provide significant benefit although it complicates the layup.

For my own experience, the particular issue I had with a keel was with an aft sweept keel. The problem stemmed from leaving the boat in a tidal creek that was not quite deep enough to prevent the keel from bottoming. The regular loading and unloading was too much for the stern most mounting points. Problem was that most of the weight of the keel was carried on the two aftmost bolts. The hull began to delaminate around the bolts and leak. I accelerated the failure but I also feel that normal use for an extended period would result in the same mode of failure.

Point here is that a keel may experience loads in any direction and the connection needs to cater for reasonable combinations. A short keel connection will experience higher stresses in a high speed grounding than a longer connection for example.

Rick W.

 

Rick,

I'm wondering if those bolts were perhaps overtightened at the repair (I asume they were not like that from origin), as the apparently bent plates and incipient corrosion suggest. This could have crushed the laminate and worsening not detected delaminations between floors and hull.

Narrow footprints require big diameter bolts which admit a big intrinsic specs torque, probably much bigger than the designed torque to hold the keel. If design torque is greatly exceeded (as it could have happened at the repair), this may lead to extensive laminate damage, specially if there is already a hidden damage after the grounding, as could have been the case. What do you think? Could this have happened?

A keel attachement like that, with no keelson, flange, socket or other way to adequately spread the loads, is perhaps an acceptable way of doing things for pure racing boats in the pursuit of weight savings, but it is not good enough engineering for a cruiser-racer, as the Cape Fear 38 is, in my humble opinion.

If I were the owner of such a kind of boat, I would be concerned about possible hidden damages even not having suffered groundings, specially if retightening of the bolts without knowing the design torques has been done.

Not to speak if groundings have happened. I've surveyed my share of hanging loose keels and other delicacies.

Cheers.

 

Guillermo
The only obvious aspect of the bolt studs is the variation in the exposed threads. Whether by design or manufacture the lack of exposed threads on some of the bolts is poor engineering.

The studs would certainly have the capacity to cause localised compressive failure of the laminate if overtorqued but it appears the laminate has failed along the edge of the keel not from the bolt holes. The integrity of a butt connection like this relies heavily on the bolt tension being adequate to avoid separation as noted above.

Would still like to see an FEM result of the connection using appropriate material properties. Must be someone here that can do this.

Rick W.

 

I went to the ProBoatbuilder article referenced here (Aug/Sept 2005) and found the highlighted text attached:

 

I'm sure there will be at least 2 people doing this sort of study. One will be the expert witness for the defence and the other the expert witness for the plaintiff in the upcoming litigation.

 

However, the edge of the keel, or flange, still bears on a surface in the same way it would on a flat surface, except in a sump-type constuction, where your example might be correct.

One major thing to consider in these design projects is the ultimate dependence on secondary bonds. Prep for secondary bonds is critical. Design of the secondary bond is critical.

Another issue during thick laminate build up is the exothermic temperature. This can have a degrading effect on the matrix.

 

Paul

I have and do design keel attachments so I do speak from experience. I do not believe any engineer was involved in that particular design and I look forward to some facts and scantling info. For the moment it is going to be more informative to illustrate why this one failed.

There is no way I will give you a simple answer for a complex design. There are several sensible ways of distributing the stresses into the hull, floors, transverse girders, longitudinals, even right to the deckhead and the solutions are as varied as the hull designs they are offered up to. If you have seen a particular attempt fail that is because it was not correctly implemented, not that it is not a viable solution.

For ideas of trends towards more robust attachment you could look at the modern large blue water endurance racers.

As for embedded metals within a composite lay-up; you will find bronze, stainless steel, and monel bedded in fiber reinforced epoxy resins within the lay-up serving very well as attachments for bolt on frames, floors, steps, chain-plates and machinery attachment points. Both in FRP and CF . But expensive since quality is important and this does not suit the cheaper building methods that make FRP so attractive.

All design of critical structures should be aimed at mitigating catastrophic failure. Composites are prone from fatigue, accidental damage and manufacturing defects to a number of failure types. The worrying ones are sub critical cracks in highly stressed areas and I don't mean micro cracking which is common and tends to terminate on a fibre but short sub critical cracks and delaminations that will grow to inevitable failure with repeated cyclic loading .

Once the crack starts if you are lucky a leak is noticed and it often is found after a club race when the vessel is back on its mooring but if the final act is in a seaway and someone dies....then it is taken more seriously and then it is news.

Some recent suggestions I have heard are that the ABS strut keel attachment area laminates should be beefed up by a factor of 4 to better insure against failures caused by hidden damage sustained in grounding.



Rick and all
Some thoughts
Structurally you proportion your thicknesses to match stiffness between dissimilar materials for a ‘gentle landing’. With backing plates between floors the floors change the stiffness like bolting to the flange of an I beam. Then the sharp corners of the main backing plate can act like the old juice can openers on the weaker material.


Consider the boat heeling and the keel moment being transferred through the fixture. The lower edge of the strut top is in compression and this is distributed along most of the interface, if we follow the stress path through to the other end of the attachment lever arm most of it passes through the farther edge of the widest backing plate. This plate being significantly wider resists almost the entire moment at its outer portion which it imparts in shear to the keel base laminate Which reinforced with floors is too stiff to yield enough to spread the load to the narrower aft plate.

All my FEA officially ends at the composite, I can model the grp as a lumped parameter isotropic material and just match E. That gives the resultants so long as things remain linear and is a good indicator, but I really don't think this needs any analysis.

Sometimes it is useful to forensically examine and fill in the pieces as they become available since when it gets to the courts the report is often unavailable in full and the issues are easily obfuscated by the parties involved.


cheers

 

Would be interesting to compare the two results. I can imagine how the model parameters will be tweaked to get the answer to support the opposing positions.

I still believe comparative FEM of the simple butt joint against a recess or stub using lead and composite properties would be enlightening. Has anyone done this?

Rick W