Bolts for a Fidanza Flywheel

V12-Engine
21

The dowels ensure exact alignment which you don’t get unless
you have fitted bolts…which are not used here. So the dowels ensure
concentricity…

Concentricity is ensured by the pilot diameter. As far as rotational position is
concerned, it was not important before the Marelli ignition system and might
not be important after. Either way, it wasn’t a factor when the decision was
made to include two dowels here.

Aluminium is
quite soft I would hope that the Fidanza Engineers calculated the
bearing load from the shear pin at failure to make sure the alloy
would hold it.

We can only hope. In the all-steel assembly, the dowels would fail in shear
well before the flywheel itself, but with steel dowels in an aluminum flywheel
the flywheel itself could deform and possibly crack.

I am surprised to hear that an
aircraft engine doesn’t make use of the friction grip. It’s available
for free and reduces the amount of components and therefore
weight…

I think our competitors at GE might have been beginning to include friction
in the calculations. P&WA didn’t, and I agree with them on that. Including
the friction makes everything sound better, but to do it right you’ve got to
calculate every possible thermal expansion and tension loading scenario to
make very sure the compression doesn’t go away right when you need it. Jet
engines simply have too many operating conditions and too much at stake (a
pilot’s life) to take such shortcuts.

– Kirbert

22

With all the information I received, I decided to check all measurements relted to the Fidanza and standard flywheel.
The thickness of the flange on the Fidanza measures: 9.65mm (49/128")
The standard flywheel: 8mm (5/16")
The tab washer thickness: 1mm
The Crank flange: 12.58mm
This tell me that I need bolts that are 9.65+1+12.58 = 23.23mm to reach through the crank flange and have the best grip.
The dowels protrude 6.9mm, so about half through the crank flange.
The clearance between the bolt heads and the “highest” point on the clutch facing the flywheel is 4.0mm. The clutch friction material before using the rivets as friction material is about 2mm.
Conclusions based on the information shared:
I need longer bolts.
I need longer dowels.
Whether I can use grade 8 bolts or not depends on the deshiphering of the codes on the flywheel bolts. Does anyone know what this mean ?

Flywheel bolt.JPG805×752 205 KB

23

The clearance between the bolt heads and the “highest” point
on the clutch facing the flywheel is 4.0mm. The clutch friction
material before using the rivets as friction material is about 2mm.

Just engaging the clutch will probably eat up about a mm of that clearance,
as the wavy spring between the friction linings is crushed flat. Still, it sounds
like you have a grip on that particular interference issue.

Conclusions based on the information shared: I need longer bolts.

There’s no need for the bolts to go all the way through the crank flange. In
fact, it might be problematic if they do, they might rub on something. You do
need at least one bolt diameter of engagement, though.

I
need longer dowels.

I didn’t understand what you said about those. Are your dowels protruding
from the back side of the flywheel? Are you using flywheel dowels or
flexplate dowels? The flexplate dowels are too short for sure, but the OEM
flywheel dowels might be OK. It only needs to protrude through the flywheel
far enough to securely engage the openings in the crank flange, perhaps
2-3mm.

And like I said, one idea would be to cut recesses in the holes so the heads
of the dowels sit down inside the flywheel and under the lock plate.

– Kirbert

24

Ok. I respectfully disagree that the flywheel is designed to operate in shear from the dowels.

From Vehicular Engine Design 2015 by Hoag & Dondlinger Chapter 16 (Cranktrain)

"The flywheel connection relies on friction to generate the shear
torque capacity in the joint, but a frictionless condition must also be designed for. "

“If a design is complete and in production, and more torque capacity is required of
the joint, various treatments can be added to increase the capacity of the joint at an
additional cost. The coefficient of friction at the joint surface can be increase by changing
the machined surface finish, by adding abrasive coatings, or by adding additional locating
dowels. Alternately, the fastener grade may be increased to allow greater clampload.”

Torque Capacity of the flange is given by :
Tc = u * Tf/(k*d) * rp * n

Where u = coefficient of friction (Flywheel to crank)
Tf = Fastener Torque
k = nut factor = 0.18 - 0.22 usually assumed 0.2
d = nominal dia of clamp bolt
rp = pitch circle radius of fasteners
n = number of fasteners

In our case, we have 65 ftlb on a 7/16" bolt. 10 bolts. 1 13/16" radius. Assuming 0.15 for the coefficient of friction (which is apparently a safe value in production), then we get 395 ft lb … which is comfortably greater than the 295 ft lb rating of the standard engine. So this results in a factor of safety of 1.3.

This is the primary operating condition of the crank - flywheel flange.

In the case where we have a frictionless condition (a failure has occurred), there are two scenarios recommended :

  1. Shear is taken on half the bolts. (Min safety factor allowed = 2)
    = T/(rp * n * Af)
    where T is engine torque, rp is the pitch circle radius of the bolts in shear, n the number of bolts (in this case 5) and Af is the cross sectional area of a single fastener (assumes shear is directed at the unthreaded bolt shank, no on the thread)

Assuming 295 ftlb max engine torque, I make this to result in 2600 psi per fastener.

  1. Or Shear is taken by one bolt or dowel (Min safety factor allowed = 1)
    Results in 13,000 psi on bolt or dowel (using formula above)
    Results in a safety factor of 4 (which seems quite high).

(I calc the max allowable shear stress on one 7/16" bolt as being ~52,000 psi.)

I don’t guarantee these numbers. I might have done something dumb. Someone can check my maths - I’m not competent with lbs and ft and things! However it points you in the direction of what is happening and what to look for.

Main issue for me is if you want to tweak an engine and deliver 571 ft lb (which I am) then the 395 ft lb capability of the standard 10 bolt arrangement with a 0.15 coefficient of friction is inadequate. The crank to flywheel interface has to have a coefficient of friction of at least 0.5 (with no factor of safety) or 0.65 to maintain at least a 1.3 factor of safety. Given an ali flywheel has potentially a lower coefficient of friction to the steel crank flange than a steel flywheel, this is another reason why alloy flywheels might not be the best solution when a steel flywheel can be made stronger with the same inertia.

Hope this helps someone.
Cheers
Mark

1 Like
25

The bolts that I have are 17.46mm (11/16") long. With 10.65mm taken by the Fidanza and lock tab plate, there are 9.8mm left to grab the crank flange. The crank flange is 12.58mm so the bolt is reaching about 3/4 of the flange. I’d like it to be all the way. There are about 2mm space behind the crank flange before a bolt would interfere with the engine block.

I didn’t understand what you said about those. Are your dowels protruding
from the back side of the flywheel? Are you using flywheel dowels or
flexplate dowels? The flexplate dowels are too short for sure, but the OEM
flywheel dowels might be OK. It only needs to protrude through the flywheel
far enough to securely engage the openings in the crank flange, perhaps
2-3mm.

Kirbert, I just went to double check. These are indeed the flywheel dowels, the others are still in the flexplate. The dowels are mounted in the Fidanza flywheel where they are recessed in the flywheel as you also suggests. They are actually protruding 11.48 !!! (I misread my own notes and got it wrong with the 6.9mm). In other words, they reach almost all the way through the crank flange. I checked the protrusion of the dowels in the standard flywheel for comparison, here they protrude 12.75mm. I’m ok with the dowels.

Thank you for all the inputs, I got a few things sorted out and can move on with my project.

26

Thanks Mark, I’ll keep this thread in mind if I ever get to do the engine upgrades that I’d ultimately like to do.

27

Ok. I think I screwed up.

Just in case anyone is still following this :dizzy_face: I am guilty of seeing a number I expected and not checking my calcs.

I am pretty sure (I have calculated it twice now with 2 different techniques) that the friction force in the flywheel joint is actually 2066 ft lbs (or 2049 ft lbs by the 2nd method) which is about 2780 Nm. (finger trouble entering the formula into my spreadsheet)

Also, the torque presented by the crank to the flywheel is not simply the mean engine torque.

In a single cylinder engine, the torque varies from -5 times to +15 times the mean torque! However, for a smooth even firing V12, the torque is varying from 60% to 140% of mean. So the peak torque on a factory standard V12 Jag engine should be about 295 ft lbs * 1.40 = 413 ft lbs. (568 Nm)

Therefore the actual factor of safety is a very healthy 5. That is, the friction of the clamped faces can withstand 5 times the peak engine torque before fretting occurs or shear is applied to the dowels/bolts.

(The factor of safety if the bolts come loose and you are relying on 5 in shear reduces to 14 and if you are relying on your dowels only to 2.8) Seems plenty of capacity left over to double the engine torque. :flushed:

Kirbert, from what I can find out, if the bolts are a tight fit (reamed) then they will operate in shear and the surface friction can probably be ignored. If they are a loose fit, then the mating faces are expected to act in friction, prior to the bolts operating in shear (as fail safe). It would be my uneducated guess that a crankshaft with oscillating torque and harmonic vibrations might be a bad thing for bolts in shear (and produce fretting?), whereas a turbine must surely have nice smooth torque and nil vibration … which was probably designed to be taken up on all the dowels in any event??

28

It is interesting to me that the peak torque on the 6 cylinder is about 2.55 x average (on the V12 it is ~ 1.42) The instantaneous torque on the 6 also goes negative (about -40% to +255%).

Therefore the same flange design, as used on the XK engine had factors of safety in a more “normal” range.

In switching to the V12, with its more even and forgiving torque, yet retaining the same flywheel arrangement, the flywheel flange actually became significantly over engineered … and should comfortably be able to handle twice the engine output.

1 Like
29

Yes…that sounds a bit more comfortable. Factor of safety of 5 . But aluminium on steel is a nice sticky connection…higher coefficient than steel on steel…but with grease and or oil there probably not much difference.
I like the textbook statement…" If a design is complete and in production and more torque capacity is required of the joint…" Heh heh…is this a polite way of pointing out a way to fix a “whoopsie…”!!!
Regards
Matt

30

I thought alloy on steel would be stickier due to greater conformity to the machined steel surface, albeit presumably a swirl pattern of near concentric face cut. The same alloy deformation applies under the bolt head, however, which may be progressive as well as instant at time of tightening. My used Fidanza has toroidal depressions maybe 1-2 thou deep around each bolt hole. I think I’ll try for a thicker steel spacer like the flexplate one and not just the lock washer plate.

From memory, the clutch dowel holes show slght edge rounding at the surface of the flywheel on the load side. However, I would guess that maximum load could be orders of magnitude higher under shock conditions either forward or reverse under brutal gear changing, and a V12 would have more rotating mass to affect the drive train load under acceleration or engine braking? Steady state torque would be very benign compared to shock loads, even with spring or polymer cush drive components.

I remember ‘designing’ a belt primary drive for my Norton before such things were on sale. Researching chain and the new HTD toothed belt specs made me realise how margins of safety were applied/ignored depending on intended uae. Standard fitment on unit-construction bikes was to use a primary chsin to connect the crank to the clutch and the chain was exactly like the XK timing chsin. Looking in power drive catalogues such chsin was rated at maybe 15hp /2500 rpm and a minimum driving sprocket size (from memory) but the engines typically developed 45 hp. and ran up to 6500 with smaller sprockets. To stay in spec for my tuned Norton I should have used a toothed belt about six inches wide, yet I knew it would be feasible to fit half that width. I think ‘official’ ratings like 12-15 hp for a duplex chain are for continuous use and low maintenance/long lifespans. You can overload them significantly

31

Kirbert, from what I can find out, if the bolts are a tight fit
(reamed) then they will operate in shear and the surface friction can
probably be ignored.

In general, bolts are not a tight fit. In fact, I think anyone who’s fiddled with
either the V12 flywheel or even the flexplate can attest that the dowels are a
tight fit (there’s a threaded hole in the middle of each to facilitate pulling them
out) while the bolts are a slip fit.

This got even weirder on the P&WA jet engines. The bolts holding rotor
disks together typically had smooth shanks plus some necked areas, so they
were thicker in some places than others. It’s been more than 30 years and I
don’t recall too perzackly where the necked areas were, but I suspect they
were right across the joints so that these bolts deliberately did NOT carry any
shear loads!

If they are a loose fit, then the mating faces
are expected to act in friction, prior to the bolts operating in shear
(as fail safe).

I’m pretty sure P&WA considered both the friction and the bolts as fail safe.
They knew it was there – how could you not? – but deliberately chose to
design with the dowels handling 100% of the torque.

It would be my uneducated guess that a crankshaft
with oscillating torque and harmonic vibrations might be a bad thing
for bolts in shear (and produce fretting?), whereas a turbine must
surely have nice smooth torque and nil vibration …

In an automotive application, you’ve got more things to consider. What
happens when the driver revs it up and dumps the clutch? What happens
when a rear wheel hits a pothole? The worst things the jet engine has to
worry about are bird strikes and ice ingestion. But, really, I think either
situation makes the case for why the dowels should be designed to take
100% of the torque, leaving the friction and bolts to handle these extreme
situations.

I also suspect that an aftermarket aluminum flywheel would inherently be
more subject to damage by misuse. It’s also probable that the designers
were not as careful about calculating shear loads in dowels, etc.

– Kirbert

32

I thought alloy on steel would be stickier due to greater conformity to the machined steel surface, albeit presumably a swirl pattern of near concentric face cut. The same alloy deformation applies under the bolt head, however, which may be progressive as well as instant at time of tightening. My used Fidanza has toroidal depressions maybe 1-2 thou deep around each bolt hole

Pete
Just a property of the material. Al v Fe is nicely higher than Fe v Fe. You can affect the coefficient of friction markedly with surface finish…and a light garnet blast makes a dramatic difference.
I think the depressions in the flywheel are a sign of over tightening. The material should be kept in the elastic range I would have thought but indentations indicate otherwise. Aluminium is about 1/3 the stiffness of steel…so flex is always a problem with alloy bits and pieces.
You’re right about chain drives. If you are designing a conveyor drive system in use 20 hrs a day then you adhere to the low side of the manufacturers load charts to get 15-20,000 hours of use…still only a couple of years…but on an old British bike I imagine they ran for 20-30,000 miles before overhaul(???) which at an average speed of …say 20 mph is only 1000 hrs…practically brand new!!!
Regards
Matt

33

Nice forum thread…

34

Just throw in another thought here.

I was running through an analysis on the head bolts (studs) to see what safety factor was designed in, when I came to an analysis of what is called “Embedment Design Margin”. This is the amount the clamping material will deform (crush) under torque from the bolt … plus thermal effects etc.

Long story short, the bolt on the flywheel should be clamping that flange to the crank. BUT … if the Ali flywheel distorts (crushes) under the bolt head (which it may only do at temperature due to additional stress caused by thermal expansion and a weakening of the ali at temp), then the bolt will loose tension. That would be bad. Evidence from my flywheel and Pete’s flywheel suggest that something of this order is happening.

The design solution is to fit a wider diameter washer under the head to prevent “embedment” within the Ali flywheel. Preferably such that there is a > 1.3 safety margin at torque + torque error + temperature effects. Fitting the factory lock ring will presumably achieve this also.

As an aside, the analysis showed that in the case of my head studs, embedment would have occurred at temperature, probably causing the head gasket to leak. The reason for this was that the studs that were fitted were not necked. Therefore they would have been too stiff (strong) and not stretch adequately when the engine heated up. I “think” the factory studs are necked, but these looked to be aftermarket or locally made (in which case the steel probably wasn’t strong enough to have caused a problem!) It is a complex problem and one that has made me cautious of any aftermarket supplier where the material specification is undefined. (not that the factory parts are defined, but at least you assume they got the spec right).

The analysis checks for fatigue, thread pull, embedment and gas pressure design margins based upon critical dimensions, material specs and tightening torque.

Rgds
Mark

35

I was running through an analysis on the head bolts (studs) to see what safety factor was designed in, when I came to an analysis of what is called “Embedment Design Margin”. This is the amount the clamping material will deform (crush) under torque from the bolt … plus thermal effects etc.

Hi Mark
This implies “post elastic” design of the bolted joint. It depends on the definition used in the analysis. Every bolted joint will deflect the material under the bolt head and the nut…but if the material is permanently compressed or stretched then it could be considered deformed. In this case it is post-elastically stressed The post elastic is not normally used by Engineers …but it does allow extra margin so highly optimised bolted joints may use them. The relative elasticity of the joints and clamping system is a critical factor in the design of the joint . I am surprised that the head to block joint analysed badly. The studs ( which on my V12 were not necked from memory…) are nice and long and therefore quite elastic. There were no signs of any deformation in the clamping faces under the head nuts. Does your analysis allow for the length of the unthreaded portion of the stud and the stiffness of the head and block down to the attachment of the studs??? The head and block part would be tricky to quantify.
Curious…
Matt

36

I am surprised that the head to block joint
analysed badly. The studs ( which on my V12 were not necked from
memory…) are nice and long and therefore quite elastic. There were
no signs of any deformation in the clamping faces under the head nuts.

OTOH, I found an issue under the smaller 3/8" nuts along the outboard edge
of the head. I have long recommended replacement of the thin washers
under the nuts with washers that are both thicker (1/8") and perhaps larger
OD. On my engine, you could feel a significant difference when torquing
those nuts to spec. With the thicker washers, they felt like they should when
torquing down, but with the OEM thin washers they felt “mushy”. When
disassembled, the thin washers would come out dished and the area of the
aluminum head surrounding the hole was similarly dished.

I agree, I don’t recall the studs being necked, either the 3/8" ones or the
7/16" ones. The head studs on my 1966 Mustang 289 were necked, though.
Of course, that assembly was all cast iron.

– Kirbert

37

Hi Matt, yes the model calcs the effective overall stiffness of the bolt, including the threaded areas and the necked areas etc. The it calculates the stiffness of the head and gasket. It then goes on to work out the clamp load, the compression on the head gasket, the deflection of the head and then the effects of gas pressure and thermal expansion on all the above.

So I originally screwed up with too high a bolt torque figure. I used a typical figure for a necked head bolt. When I reduced torque to spec (70Nm) the non-necked stud works fine. Of course lots of assumptions with stud material and head material and gasket thickness etc etc.

Biggest assumption (and probably an error) is assuming 4 head studs per cylinder. In fact the V12 shares 2 studs with adjacent cylinders. So I am going to have to figure the impact of that. Is the gas pressure only relevant to the firing cylinder (which has effectively 4 head studs during this event) or is the overall clamping force reduced by a factor 14 studs over 6 cylinders. Average of 2.3 studs per cylinder!

Anyway, assuming 4 studs per cylinder, the standard 90mm bore gives numbers something like as follows :slight_smile:

Jag V12 STD Cylinder Head Gasket Sealing Limits.jpg748×515 41.7 KB

As an example of calcs:

Head stud is at 37.5% of proof stress (assuming class 12.9 fastener) which equates to 70Nm. This is very low but works with the very stiff stud.
Inital stud stretch is 0.31mm while head and gasket compresses 0.10mm (assuming 0.7mm graphite gasket).
Adding gas pressure and the stud stretches 0.33mm and the head and gasket compression reduces to 0.07mm (taking crush off the gasket).
The initial clamp load (from stud) is 35kN and this varies -8 kN (at -30°C) to +13kN (+107°C) with temperature and -8kN with the gas force. So worst case sealing at cold start is 35kN - 8kN - 8kN = 19kN (per stud).
A graphite head gasket is spec’d at 175N per linear mm or about 12.4kN required per stud in this 90mm bore application. This results in a Design Margin of 19kN / 12.4 kN = 1.5

Endurance is calculated looking at the cyclic load variation w.r.t. the endurance strength and the ultimate tensile strength for the stud material. Assuming 4340 steel, this comes out to a design margin of 2.

Suggested thread engagement is usually 2.5 x Dia. In this case I think it is only about 1.7 x Dia. However the cross sectional area of the engaged threads is large and assuming a yield strength of the Ali block of 225 MPa we get a design margin of 1.5. Looking at it another way, the model predicts pulling the studs out of the block if you tighten beyond 88 ft-lb (119Nm) … the steel stud would have stretched about 0.52mm (but be nowhere near to breaking), but the Ali threads would yield.

Finally, the stress under the head bolt from the clamping is about 148 MPa. With a yield strength of 225 MPa we have a design margin of 1.5 again before the nut/washer starts to embed itself into the head. In theory this should be calculated with a hot head, which will reduce the yield strength, but I don’t know by how much.

This has been an interesting exercise for me. Hopefully I can use this model to look at the main bearing cap studs, which I am a bit worried about in terms of thread engagement in a high power application. I had a question in my mind of whether TWR helicoiled the threads.

Note: There are a lot of reports of power limited by head gasket sealing capability. If I recalculate this with less than 4 studs per cylinder (which no doubt I should do) the sealing margin reduces significantly (ie fails!) but this can be offset by increased bolt torque or different gasket technology (embossed gaskets). I need to have a head scratch about that.

1 Like
38

Biggest assumption (and probably an error) is assuming 4 head studs
per cylinder.

Are you just ignoring the smaller 3/8" studs?

– Kirbert

39

Yes.

It was easier that way!

Looking at it, I suspect they are more to do with sealing coolant etc, rather than providing cylinder clamping.

But I fear the only way to do it properly is to treat the entire head as a clamp load, with all the studs accounted for. That’s going to mess with my head, but I might give it a go.

40

Hi Mark
I think your assumption for clamping force resisted by 4 bolts is valid. The adjacent cylinders are not firing at the same time so the clamping force will be supplied by the 4 bolts. I imagine the cylinder head experiences a sequence of compression events along its length and the shared bolts will get twice the load events…one from each adjacent cylinder…but fatigue is not an issue.
An interesting learning is the lower margin on sealing on a cold engine. …19kN vs 40kN…Adds another powerful reason NOT to thrash a cold engine…
The fatigue life is a function of mean stress and the magnitude of the fluctuating stress so I would hope your analysis incorporates these parameters somewhere in the numbers.
Very interesting stuff. Have you modelled the cylinder head with all its cooling channels other features…?? That is a lot of work without original drawings!!!
Respect…and regards…
Matt