How much pressure does driving a rivet exert on its hole? 2

[KirbyWan]

So I'm looking at repairing a small access door using a doubler, but the door is plastic and I was wondering if the pressure from driving or squeezing a rivet might cause the door to crack. The door material is .150 thick epoxy sheet molding compoud per BMS8-327 Type 1 which indicates a design material strength of 30 ksi. The material spec gives a strenght of 35 ksi (Boeing being more conservative of course). Is there a reference I can use to calculate the rivet force on the internal face of its hole from being installed?

I suppose this is not a problem with metal materials becuase they are ductile and expand as needed when the rivet overcomes their yield strength, but on a brittle plastic material one pound of force over F(tu) and *POP* ... toss it in the trash and get a new one. And yes I have thought about a fiberglass layup repair, but it's just not optimal for the location of this crack repair.

Thanks,

-Kirby

Kirby Wilkerson

Remember, first define the problem, then solve it.

 

[rb1957]

was there a rivet in the panel before ?

i don't think the swell is That much, but i think you're right to be concerned with the plastic. squeeze the rivet rather than driving it. maybe use a softer material (B, A, instead of the typical AD).

is a Hi-Lite unacceptable ?

 

[KirbyWan]

The original crack that I'm trying to repair extends from the open area from the hold down clip through one of the four fasteners, which is where I think the crack originated from over-tightening the fastener to the corner of the panel. I wnat to drill new holes and along the panel edge and double flush some rivets so that the panel can still close without causing an interference problem.

The problem is I want to use a SS doubler to minimise thickness in the airstream, but monel rivets would create even more stress then AD rivets. I can get enough strength with AD rivets, but then I've got dissimilar metal corrosion problems. Would installing rivets wet (which I usually do always) be enough or should I just go for the monel rivets?

-Kirby

Kirby Wilkerson

Remember, first define the problem, then solve it.

 

[rb1957]

it sounds like you're reparing a plastic panel that cracked at one of the corner fasteners. it sound like you've cut out the cracked attmt hole, or are you crack stopping?

if you're CSKing into the dblr, it'll need to be at least .050" thick ... i'd've thought that 0.05" of 2024 would have repaired 0.15" of plastic.

i would not mix AD rivets in SS dblr.

maybe install bushes in the attmt holes (to stop over-tightening) ...

 

[KirbyWan]

I was going with SS to minimize thickness. With NAS1097 head rivets (I can't remember what the callout for Monel rivets with the smaller heads is) I can go down to .040 thickness. With 2024 Ftu ~63 ksi repair BMS8-327 @30 ksi I would need .071 minimum and possibly .080 with a 15% additional saftey factor. That is a little thicker then I like, but if that's what's needed I'll go that way. Can't put anything on the inside because it will interfere with the surface the door rests on when closed.

The crack goes from the outside edge of the panel through the latch fastener hole to the inner slot where the latch release is, so there is no stop drilling. The fastener hole with the crack going through it I was going to leave as is. The doubler will keep the ends of the crack together and EA934 adhesive will fill and seal the crack.

Actually this may be purely a learning excercise as they are still working on finding a replacement door.

-Kirby

-Kirby

Kirby Wilkerson

Remember, first define the problem, then solve it.

 

[rb1957]

replacing the door sounds like the best way to go.

i'm not familiar with BMS8-327 but i'm guessing it's a full head rivet, maybe something like a 6 if you need 0.071" to hide the CSK. If you're using 6's, why not LZ6 ? but then why 6's ? how much load can be in a 0.15" thick piece of plastic ?? but then your alternative in a monel 1097-4 (if you're going to put it in 0.04", i expect it's a -4) ... sorry, back to how much load can there be in 0.15" thick plastic ? why not 0.05" Al and LZ4s

you've got a crack in the plastic that's run-out into a slot in the cover, and the dblr is bridging the crack, yes ? touch CSKing on the inner face ??

 

[KirbyWan]

The BMS8-327 is the Boeing specification for "Epoxy Sheet Moulding Compound". It is a thermoset plastic impregnated with .5" glass fibers. This is in fact an FRP, a fiberglass reinforced plastic, think of it as a more easily shapable version of BMS8-79 fiberglass cloth prepreg. The BMS spec mentioned above gives it a design strength of 30 ksi which is what I'm using as the basis for my repair design. I think you have picked up on my use of the word plastic and not on the fiberglass reinforced part (I just looked over this thread and I failed to mention it outside of calling it epoxy sheet moulding compound.)

Here's a couple of pictures:

http://files.engineering.com/getfil...-4cd7-8f4b-057e147c5328&file=EO_10-221-02.JPG

http://files.engineering.com/getfil...-46a0-86b0-b3c4fdb6e2e4&file=EO_10-221-04.JPG

-Kirby

Kirby Wilkerson

Remember, first define the problem, then solve it.

 

[rb1957]

thx for the pix ... so it'll be the screw CSKs that define the thickness of the dblr (i'd put a patch over the corner). is there enough room outside the edge stiffener to attach rivets ? i expect there's a seal along the edge.

 

[KirbyWan]

The distance from the stiffening rib to the door edge is 1/2 inch. So if they were centered I would need to go down to #4 rivets, If I used #5 rivets I would need .313 ED. With half of the CS diameter of .2976 for a #5 That would be .461 so I could fit a #5 MS20427M5 with the full countersink on the inner face and touch countersink the SS doubler.

Now I'm stuck with two problems, first my original problem of knowing whether squeezing the rivets will crack the FRP panel, second trying to figure out what the minimum joint strength is per fastener. I know the shear strength for a monel #5 rivet, but how do I get the lower value for a CS rivet in SS. I can't use the values in MMPDS because they don't give the values for a #5 in .040 thick material because that would be a knife edge condition, but by doing just a touch countersink I avoid the knife edge.

Am I correct in assuming the values in MMPDS are from testing? Or is there a mathematical relationship based on bearing area of the contersink? If a test is needed how extensive should my test be to provide a strength per fastener. Am I over thinking this?

Thanks for you help RB.

Kirby Wilkerson

Remember, first define the problem, then solve it.

 

[rb1957]

do you have a good handle on the load applied ?

since it's a plastic panel, are you going to glue the crack faces together ? you won't be able to count on it for strength, but is seems a reasonable thing to do. how about gluing the repair dblr on as well ?

is it unthinkable to relocate the latch away from the corner ?? (quite possibly ! but it would be nice)

will you be able to squeeze monel rivets, or do you have to drive them ? squeezing would be much nicer for the panel.

can you get hold of some plastic to try or it sounds to me like a pretty reasonable concern.

to me 0.05" or 0.063" Al would be enough to repair a fiberglass cover, but YJ (Your Job) ...

 

[KirbyWan]

RB,

I agree if I could get some basis for defining the loads on a panel. Without that I am limited to restoring the original strength of the panel as determined by the material used. So my analysis is going to be simply P(loss) is 1.5" crack in .150 thick material with a defined strength of 30 ksi = 6750 lbs. P(gain) for the doubler and the fasteners must meet or exceed this usually with an additional 15% margin. (I don't think I need it for the fasteners because I'm using more then 5 fasteners.) Maybe this would be a good time to toss out the question "Is this the standard practice for repair design?" Still learning this trade and perhaps being over conservative along the way.

-Kirby

Kirby Wilkerson

Remember, first define the problem, then solve it.

 

[rb1957]

i appreciate the limitations you're working under, and i accept that this type of repair design is common and may be "blessed" by AC43-13 but i think it leads to bad repairs. to be consistent, if your repair dblr is good for 6750 lbs, then you need to attach it with rivets good for 6750 lbs (well 13500 lbs as you've got to shear the load into and out of the dblr) and this doesn't look sensible (and in my experience has lead to very poor repair design, as rivets are squeezed into the limited available area).

There is no way that the panel is stressed to Ftu in service. Maybe a rational basis is to suppose that the repair is transferring the latch load, 50 lbs ? 100 lbs? (sounds horribly conservative). Maybe the load from the latches (4?) is reacted along the perimeter of the panel.

i wonder if you've got good sense as to what initiated the crack ? (not a knock, just a question) i think you thought it was over driving one of the latch screws. could it have been an "apron ape" pushing the panel closed to get the latch to engage ??

still trying to source a replacement panel ?

 

[KirbyWan]

I haven't heard the term apron ape before. That made me laugh. These latches are pretty stiff. My first guess is over torquing the latch mounting screws.

I agree that this repair is overdesigned, but to meet regulatory approval is there some other out that I have to get a lower value of what load needs to be reacted? I feel at the very least I need to be able to say to the FAA, 'here is where I got the minimum design requirements for my repair.' The load that the clamps are rated to is a good source, but that would be a flexing load tending to bend the edge of the door up. I could calculate the load that could be carried through the hinge which dosen't look that substantial. Or maybe I can just say that the corner of the door can not be a significant load path and with a bit of hand waving say only this is needed. Of course it is material that was rated at 30 ksi that then did crack. which means that it was enough force to at least start the crack which then propigated to the edge through vibration.

-Kirby

Kirby Wilkerson

Remember, first define the problem, then solve it.

 

[MNLiaison]

Values for csk fasteners in the MMPDS are from testing while those for protruding head fasteners are typically a calculated value using Fbru.

I personally would bond a doubler on using EA9394 in lieu of riveting if I had to fix the damage. Holes are always the enemy in composites and bonding provides a much easier fix for this situation. I still might provide a single row of rivets around the periphery if I was worried about the bond coming apart and the plate causing additional a/c damage down wind.

 

[KirbyWan]

I was planning on bonding it using EA934, but not use the adhesive strength for my analysis. If I'm doing a metal doubler on fiberglass I will only use fastener strengths in my analasys. I generally will not use adhesive strength unless it is a heat cured film adhesive or prepreg.

-Kirby

Kirby Wilkerson

Remember, first define the problem, then solve it.

 

[rb1957]

fwiw ... i agree, RT adhesive is there in reality but not analytically, just like gluing the crack together is a good thing to do but not worth anything in your analysis; but you're not designing the fasteners to carry 6750 lbs in and out of the dblr ? if so, why design the dblr for that load ??

does the original plastic panel carry pressure loads ? (i don't think so)

 

[KirbyWan]

No pressure loads. Just an access door on a 737 fan cowl. And yes I am designing the fasteners to carry 6750 lbs. Again that is the design requirement without some form of handwaving that says less is needed. When I design a repair doubler, if the original repair could carry a given tension load, my doubler must carry that same tension load. If my doubler can carry that tension load, then the fasteners must be able to transfer that tension load into the parent structure. Over design often happens, but I can't go less and prove my repair is valid.

Having looked at this I may go back and see if there is a way to do a fiberglass prepreg repair. It's looking better and better even given the tight location.

-Kirby

Kirby Wilkerson

Remember, first define the problem, then solve it.

 

[rb1957]

f/g p/g repair ... say 10 plies on the OML ...

 

[SWComposites]

1) why do you need to repair the part in the first place? it is probably ok as-is; though maybe hard to prove

2) the part design is bad; having a fastener hole so close to the cutout corner is poor design as there are 2 stress concentrations interacting

3) the crack likely started between the fastener hole and the cutout, then extended to the edge of the part

4) this material is not homogeneous and the fiber orientations are random and affected by flow in the mold, and there is a possibility that the part had a low strength area where the crack formed

5) "analysis is going to be simply P(loss) is 1.5" crack in .150 thick material with a defined strength of 30 ksi = 6750 lbs" - I very highly doubt there is anything near that load in the panel at that location

6) putting a metal doubler on with a bunch of fastener holes is very likely to reduce the part strength rather than improve it, particularly if the fasteners are anywhere near the part edges. Along the edges the fiber orientations are probably not isotropic, but oriented parallel to the edge (it all depends on the material flow when the part is molded), and the strength near the edge may be low. If the rivets are close to the edge you are very likely to crack the part during installation.

7) I would consider putting a couple of plies of fiberglass on with EA9390 wet layup resin and calling it good. This is probably a non-structural part, so not much in the way of analysis should be required.

SW

 

[KirbyWan]

SWComposites,

I agree with everything you said, my original intent with this question was to find out the stress that squeezing a rivet will induce on the fastener hole because I was worried about cracking the edge out. Can I say in my analysis, 'This access door does not carry structural loads and no analysis is required.' And then design the repair to what I think is good enough?

This is for a customer who will want an 8110-3 on this part. I know you are all going to say this is a minor repair and no 8110 is needed. You might even say there should be no 8110, providing one is against the policies and procedures defined by the FAA. But the customer is a foreign carrier and if the repair is not per the CMM then an EO is needed and if an EO is needed it must be accompanied by an 8110. It's wrong, but that's the way the customer wants it. This puts us between Scylla and Charybdis, but what else is new.

-Kirby

Kirby Wilkerson

Remember, first define the problem, then solve it.