Valve bounce

morrisnutter · 4 Dec 2006, 20:07 (Ref:1781949)

Does anyone know how you can calculate the revs at which valve bounce occurs by starting with measurable parameters? I thought it might simply be the natural frequency of a mass (ie that of the valve) on a spring, but I think it's more to do with the valve impacting the seat. Is it to do with the stiffness of the head (ie the valve head becomes more dished) and springing back?

Anybody know?!?!?

Thanks

Alex Hodgkinson · 4 Dec 2006, 21:19 (Ref:1781991)

I don't have an answer but I think that you'll have to consider the effect of the air entering the engine, especially with forced induction motors.
You'll also have to look at the thrust velocity of the cams (ie their profile.)

Not something I've ever calculated myself, but if I were you I'd be expecting a big sum to do!

ian_w · 5 Dec 2006, 18:56 (Ref:1782741)

This is a very complex subject. To do even a basic calculation you will need to know the acceleration profile of the cam. If you have a very accurate lift curve you can double differentiate to get acceleration, this will typically be in units of mm/deg^2 or mm/rad^2. If you convert engine speed from rpm to deg/s or rad/s you can substitute this into the acceleration figures and convert them to mm/s^2.

If you know the mass of all the moving bits of your valve train you can multiply this by the acceleration to give a force. This is slightly complicated by the design of some valvetrains as not everything moves the same amount as the valves ( i.e. cam follower on OHV engines ). If the engine uses rockers you will need to know the position of the CofG of this to calculate an effective mass for this.

Anyway, now you should have some force numbers for the cam. In order for the follower to remain in contact with the cam then the valve spring must supply at least this amount of force. Typically at the design stage you would chose a spring with 20% more load than the minimum you have calculated from the cam profile. This gives you some margin for errors and nasty things like spring surge.

I did warn you it wasn't simple.....

GordonG · 5 Dec 2006, 20:41 (Ref:1782802)

I believe you're calculating valve float there, not valve bounce.

Float is where the follower leaves the cam on the closing ramp. Bounce is when the head hits the seat so hard it bounces off the seat again, effectively leaving the valve open longer than designed and not doing anything in the area any favours.

I expect bounce will be far harder to predict as it will be to do with the velocity of the valve when it hits the seat (and this may only start happening after valve float), combined with the weights, flexibilities and resonant frequencies of the valve train, in particular the valve (head and stem), spring, seat, retainer etc. Way beyond my analytical skills.

G

johnny yuma · 6 Dec 2006, 00:28 (Ref:1782929)

Perhaps a practical solution by trial and error was arrived at historically
before anyone did the math and nobody bothered with the theory just drew up some tables as guidelines,like timber spanning tables.

an inner valve spring as I understand it is put there not so much for the added stiffness but to "interrupt" the resonance of the standard spring at certain RPM WITHOUT loading up the wearing parts too much

morrisnutter · 6 Dec 2006, 09:36 (Ref:1783133)

I think I shall go the trial and error route- just thought I'd ask to see how it's done.

What ian_w says is useful though. Many bang-ups are caused by valves staying open for too long and having a brief liason with a piston. Often attributed to the follower coming out of contact with the cam. Being able to size a spring to act on the rocker (thus the pushrod and follower) helps in reducing the likelihood of this happening.

Thanks

Paul

norman-normal · 6 Dec 2006, 16:06 (Ref:1783400)

Excuse me, but why would you want to?

morrisnutter · 6 Dec 2006, 16:37 (Ref:1783422)

Quote:

Originally Posted by norman-normal

Excuse me, but why would you want to?

If you know at what revs it's going to happen then you can select the correct springs from the very start so that you never run into bounce. It's rather the same as calculating the modes of vibration on a crankshaft. You can go the trial and error route, but it's often (though not in this case) simpler and cheaper to work it out!

norman-normal · 6 Dec 2006, 18:58 (Ref:1783507)

In that case and in answer to your question, you can't. The variables are way too complex. The valves don't all start to bounce at once, the weight of the components, the valve train,(OHC or push rod?). The best way to postpone VB would be to lighten the valve gear if possible without damaging the integrity of the parts. Stiff valve springs break parts and cause rapid wear.

What type of engine and for what purpose?

morrisnutter · 7 Dec 2006, 12:53 (Ref:1784059)

It's a 1935 Riley 12/4. 1500cc twin underhead cam with pushrod to hemispherical chambers. Planning to use it as a hot race/sprint engine revving quite hard.

I'll go the iteration route I think!

Chris Roden · 7 Dec 2006, 13:11 (Ref:1784083)

Paul
You have obviously got too much time on your hands mate!!

Talk to Keith Pointing (in the members list), a very helpful chap.

I have run double valve springs in mine and had bounce at about 6500 rpm.

When I changed over to the alternate head, just before Donington, Keith sent me some single springs that are much shorter than standard (approx 3/4"), and feel much less "strong", along with some aluminium spacers.

I no longer have a valve bounce problem.

phoenix · 7 Dec 2006, 15:12 (Ref:1784206)

Quote:

Originally Posted by morrisnutter

It's a 1935 Riley 12/4. 1500cc twin underhead cam with pushrod to hemispherical chambers. Planning to use it as a hot race/sprint engine revving quite hard.

I'll go the iteration route I think!

I too can't help you with your calculations. As has been said already, cam profiles and attributes are required to even begin the calculations.

However, if you could give cam timing and lift I could give you an approximate estimate of peak cam follower acceleration/deceleration - you could work out the loads at the spring by weighing all the valve train components and multiplying them by the acceleration/deceleration and the rocker ratio.

I would think that the rev limiting factor for this engine would be the 100mm stroke - not valve bounce. Peak power as standard was at 5500 rpm, if I'm not mistaken, giving a mean piston speed of 18 metres per second - thats the same piston speed as a YB Cosworth engine at 7100 RPM! Mean piston speeds of over 20 metres per second result in engines that don't last very long. For example, V8 F1 engines next year will have mean piston speeds of 25 metres per second at 19000 rpm - but they get a lot of new components every rebuild.

I don't know how long the con rods are, but I suspect about 6.5" or more, so they will be quite heavy, so limiting safe maximum RPM as there will be huge inertia loads on the crank, the bigend cap and bolts and the little end of the rod, loads which rise exponentially with engine speed.

However, the engine has very short push rods. Substituting a lighter material for the pushrods and taking some material from the valves will save some, but not much, valve train weight, thus hopefully allowing the standard valve springs to cope with a few more revs. If the rods and crank don't give in - or are modified to cope together with lighter pistons, I should think 6000 rpm (mean piston speed of 20 metres per second), would be achievable and I doubt that valve bounce would be a problem at that speed.

Good luck!

norman-normal · 7 Dec 2006, 16:45 (Ref:1784253)

Morris, I believe overhauled one of these about 50yrs ago , does this engine have aluminum connecting rods as standard equipment? At any rate, it is a very long stroke engine and totally unsuited to RPM's of 5,000 or more, as Phoenix stated so well.
As I see it, you best hope for improvement would be: raise the compression (the hemi head should tolerate 10.5to1) and stay with the stock cam or one with greater lift and duration but the same timing.