What's new
Van's Air Force

Don't miss anything! Register now for full access to the definitive RV support community.

Viking 110 notice

I have been working on a new flywheel proposal, based loosely around what Aeromomentum uses on their Suzuki engines. They did not provide me with a print so I am just going with something that is not too complicated to make, while keeping as much mass as possible out at the rim for best angular inertia vs total weight. I started out looking at Honda racing flywheels, but they are more focused on reducing mass and reducing rotating inertia - so not the same goal. But I did gather that their flywheels got down to about 8lb with the mass centered on the clutch friction face.

So far the weight is about 9lb (I have not simulated the teeth on the ring gear so its slightly heavier). The crankshaft flange area is 8mm and then the web thins to 6mm with the rim being 18mm wide. The OD of the ring gear is 11.125" 282.56mm.

When looking at the stress from only the RPM (centrifugal load) the peak stress is right around 100MPa in some small areas near the smaller lightening holes. So far so good. I then did a calculation for the gyroscopic torque reaction with the flywheel rotating at 5800rpm and yawing at 1 turn per second. With a mass of 4.08kg, a radius of gyration of 0.12m, rotational speed of 607.45 radians per second around x and 2xpi around z the torsional reaction works out to 196Nm.

Since I have discovered that I cant use 2 centrifugal constraints in the FreeCAD FEA I had to come up with another method to simulate the gyroscopic torque reaction. So what I did was to apply 2 forces direct to the rim of the flywheel 180 degrees apart and in opposite directions to create the torque couple. The loads were 819N each at a radius of 0.12m.

This ran successfully in the solver and has the appropriate effect of wanting to deform the flywheel web around the hub (a rotating bending moment) which is exactly what led to the flex plates cracking in the way that they did.
I might be inclined to taper the 8mm web out at a shallow angle into the 6mm surrounding web to reduce any stress concentration. These calculations have no factor of safety applied. I dont know how often people exceed the redline in a dive, for instance. The engines are not aerobatic either so hopefully a turn at 1 revolution per second is fast enough... Peak stress shows as 158MPa which is a safe "forever load" for a good quality piece of steel with a suitable surface finish. More than likely that finish will be protected by some good quality paint.
 

Attachments

  • Flywheel-2 result.JPG
    Flywheel-2 result.JPG
    89.2 KB · Views: 80
  • Flywheel-3 gyroscopic result.JPG
    Flywheel-3 gyroscopic result.JPG
    129.2 KB · Views: 91
  • Flywheel-4 gyroscopic result.JPG
    Flywheel-4 gyroscopic result.JPG
    88.6 KB · Views: 94
  • Flywheel-5 gyroscopic result.JPG
    Flywheel-5 gyroscopic result.JPG
    63.7 KB · Views: 62
Last edited:
I ended up running the simulation of the flex plate without the lightening holes, but for the gyroscopic forces I used the original mass from the model with all the holes, 1.6kg.

The gyroscopic torque moment for 5800rpm and 1 turn per second worked out to be 79.36Nm, much less than the billet flywheel. However the stress level at the outer edge of the crankshaft flange was just a hair under 400Mpa.

Any time one is turning there is a bending load applied at that attachment point and load reversals occur at engine rpm or 96.66hz at 5800rpm. So one will accumulate cycles very fast in practice. By 1 million cycles the material SN curve would already be below 400Mpa so the fact that the flexplate would be severely cracked by 350 hours is not a big surprise.

The rim of the flex plate is deflecting 0.9mm due to the gyroscopic loads based just off the weight of the ring gear itself. If the lightning holes were added overall stress levels would be higher but I just could not get the mesh to behave with the system that I had.
 

Attachments

  • flex-plate-01.JPG
    flex-plate-01.JPG
    118.9 KB · Views: 96
  • flex-plate-02.JPG
    flex-plate-02.JPG
    121.7 KB · Views: 75
  • flex-plate-03.JPG
    flex-plate-03.JPG
    72.8 KB · Views: 65
  • flex-plate-04.JPG
    flex-plate-04.JPG
    75.8 KB · Views: 71
Removing the ring gear that Viking added to the flex plate and re-analysing. The mass dropped from 1.6kg to 1.12kg. Without lightening holes. Applying the new weight, the gyroscopic torque value for 1 turn per second yaw is now 39.8Nm. I applied this as 2 point loads of 147.5N each at a radius of 135mm in 2 opposite directions.

The resulting peak stress at the outer edge of the crankshaft is 243MPa where the first bend is in the flex plate center. 243MPa is getting up there for a "safe stress" but then the amount of time that a car (with an automatic transmission) actually spends at 5800rpm while doing 1 turn per second is pretty low. But it also suggests that this part has been designed with very little design margin in the application. Not enough to allow welding a ring gear to it, Like what Viking is want to do.
 

Attachments

  • flex-plate-05.JPG
    flex-plate-05.JPG
    146.4 KB · Views: 83
Keith,
I really wish I understood the science behind what you are doing but sure glad you are doing it. Alternate engines are better for all of aviation as long as they do what they are supposed to do! Thanks - I will have to come on down and see you one of these days.
 
Keith, a side note...

The pitch or yaw rate of 6.28 radians/sec (one rev per second) is Pitts spin territory, so some observers may buck at the assumption. Not me, but some. The 1 rad/sec I previously proposed would be the dead minimum.

It does bring a detail to mind. Viking installations tend to lack motor mount triangulation. John said the cracked flexplates were limited to the 110 installations. Do those 110's lack mount stiffness?
 
I dont have a motor mount to examine, so I cant really comment. I wouldnt be so certain that the flex plate issue is limited to the 110 only. Jan is doing the exact same thing to the 130 and 150 hp motors and there is no substantial difference to the flex plate. I happen to have a virgin R18 flex plate in my possession and it is the same material thickness as on the Fit engine, just ever so slightly bigger in diameter. It has 8 bolt holes for the torque converter instead of 6, same at the crankshaft.
 
If one was absolutely trying to avoid a flywheel (and Im not sure why exactly one would want to do that) one could have something like 4140 sheet 4mm thick water jet cut to make flex plates out of. If one wanted to do it right, just a single forming operation to form a flange of the right diameter for the ring gear. Chucking off the OD, cut the pilot hole for the crank, now locating off the pilot hole true up the OD to a suitable fit to accept the ring gear. Should be pretty inexpensive and not very heavy but with a 4mm web it would be capable of holding a ring gear without failing compared to the 2.5mm material in the Honda part.

However, I think that taming the engine pulsations a bit on such a high compression engine will help when trying to resolve the TV issues "down the road" which we will come to soon enough. The flywheel concept I have is roughly 9lb. When we got to looking at the TV study one could start with the full weight flywheel and then in succession, shave more and more material off the rim to reduce the weight and see the impact on the TV sweep. One may have to throw away the prototype flywheel when you are done, but now you know the trade-off of flywheel mass vs TV input / output through the Centaflex coupling and can make the right decision regarding where to compromise.
 
I have had people say that from a design point of view I should be running 7000rpm instead of 5800rpm because apparently the rev limiter just starts kicking in at 6800 in a descent. If one goes too crazy one would end up with something too heavy. The gyroscopic loads mainly influence the attachment point of the flywheel and the thickness of the web. Striving for a high rotational inertia is mainly a function of how much mass you pack into the rim under (and next to) the ring gear. The less mass in the rim, the thinner you can go with the web, to a point. One must at the very least be able to carry the ring gear.

The historical giubo that has been used in countless applications has been the 95mm pcd BMW part. The flywheel bolts are on a 63mm pcd which puts the attachment points of the giubo very close to the flywheel bolts. Because of the tender nature of the flex plate, the flywheel drive flange assembly was added which is made very strong. It and the giubo weighs 2.6lb. The flex plate with ring gear 3.5lb. But the weight of the giubo and drive flange is near the centerline thus not contributing any to damping the engine excitation.

The new coupling has an OD of 150mm 6" and bolt attachment PCD of 125mm. I like the fact that the attachment bolts are further away from the crankshaft itself and the web is going to be thick enough for the area to not be overstressed.
 
I believe the flexplate/ring gear cracks are a sperate issue from the gearbox drive coupler failures. The flexplate is a stock Honda part with a starter ring gear welded to it. The welds are as good as any I have seen, they are uniform, and the plate shows no sign of warpage. There are no additional holes drilled.

The cracks appear to be limited to the 110 engines and not the 3-gear setup on the 130/150/190 engine which use the same part. Charlie Rosenzweig has the 3-gear version with the same flexplate/ring gear on his turbo 1.8L Honda, so it would be interesting to see if he has the same problem. If the cause is pitch/yaw induced bending it would affect every installation of this part.

So what is different about the 110 setup that could cause a bending vibration? The big difference is the bottom half of the flexplate is in the radiator airstream where it is subject to the prop pressure pulses. It may be the same phenomenon at work as cracked exhaust pipe-muffler joints because the pipe sticks out a bit too far from the cowling. The other engine installations use a horizontal radiator mounted under the footwells, vs the 45 deg angled 110's which is about 10" behind the prop. It could also be turbulent flow through all those holes are causing some resonance to occur.

Just one more thing to check during the CI.

John Salak
RV-12 N896HS

I just stumbled upon this post. I do not have a viking supplied flex plate. I made my own, by welding a ring gear on a flexplate. I do agree this is far from ideal, IMHO, but I did not see another option at the time I did this, which is probably more than 5 years ago now. I do have a 3 gear Viking supplied gearbox, But nothing else in my set up was supplied by Viking. I have inspected my flexplate carefully and I do not see any cracking. I also do not have any measurable end play in my GB input shaft.
 
Charlie, did you unbolt it for the inspection and remove it from the crankshaft ? Because if you didn't you would not be able to see the cracks. The cracks are from the web being too thin for the gyroscopic load of the added ring gear that it was not designed for.
 
Keith,

I have not. I have some significant non-aviation things going on now that I have to work on for a while. My flexplate is quite a bit different than the ones used on the 1.5L 110 set ups.

Charlie
 
Back
Top