Exhaust valve bounce

[Kevin Johnson]

You are using "weights" (plural) which implies changing more than one specific type or class of component in the combination. So really you are not changing one thing and holding everything else the same. Even holding the mass constant in a single component but changing the mass distribution can affect the resonant frequency.

I'm defining the reciprocating mass as a single mass in this thought experiment and including half the spring weight with it. Do you have any reason to believe that this is a quantitatively poor approximation for this problem in a direct acting bucket DOHC valvetrain?

Just randomly: reducing the mass of the bucket lifter neglects that it is rotating.

https://www.google.com/search?q=kinetic ... ating+body

[Rick!]

The problem is how do you account for EVO and remaining cylinder pressure and that changes with VE and rpm. That with added deflection induces un calculated valve velocity.

I’m not worried about EVO event in terms of exhaust valve bounce. The pressure is higher in the cylinder than in the port, the valve has been closed for a while, and the camshaft is trying to open the valve. I think that the valve and lifter are going to stay on the cam lobe reasonably well at that point. This is a DOHC with hydraulic direct acting bucket lifters, so I am thinking it’s pretty rigid and it’s not going to spring load the valvetrain the same way as a pushrod system. Is that a safe assumption?

I’m specifically worried about the EVC and valve bounce related to that event.

List out the components in the valvetrain and determine where your assumption is valid and where it might not be.
Is the cam really stiff or does it flex?
How thick is the bucket face? How thin is it where the cam first contacts it?
How many bits between the bucket face and the valve pad? (EDM a lifter bucket to look at its guts)
Bulk modulus of the oil?
Stiffness of the spring retainer?

All these stiffnesses are in series so the net stiffness is softer than the stiffest component in the system.
You also mentioned increasing rpm. So, F/a= dynamic mass. This comes in handy if making lumped mass estimations due to a lack of component inertia properties.
Match your new dynamic mass with the current dynamic mass and see if your linear assumption is proper.
Tangential acceleration increase will probably be a pretty significant factor in determining spring seat pressure.
You allude to knowing what variable will trigger exhaust valve float in the current system. How do you know that?
Did you just keep turning up boost until it you lost power on the chassis dyno? Was it exhaust valve float, lifters or ignition?
What's the drive pressure at the higher boost level?

Anyway, figuring out your requirements from two directions and comparing results is better than only looking at one method, in my experience.

[ptuomov]

IVO to mid-lift valve motion is controlled by the camshaft lobe. Mid-lift to max to mid-lift is controlled by spring force.

Spring force is a function of spring rate, installed height and valve lift. Since valve lift changes with crank angle, so do spring forces;
they peak at max valve lift.

Valve train force is a function of ensemble mass (valve, retainer, keeper, spring, push rod, rocker, tappet, etc),
and valve acceleration. Valve acceleration is determined by cam lobe shape and rpm.

Now things get interesting. Both spring force and valve forces will change with crank angle; valve forces are also
affected by engine rpm.

What is necessary is to plot both force curves against crank angle (at various engine rpm), to find the engine rpm at
which valve acceleration force exceed the spring control force. At the cross-over point, the valve ensemble will separate
from the cam lobe and the valve will float over the nose of the cam. When the valve floats over the nose, it crashes
down on the seat at closing, causing it to bounce. And bounce it does, often quite high off the seat.

The valve separation point, where spring control is lost, does not occur at the nose or at peak lift. It occurs at peak
valve acceleration crank angle. Knowing the actual cam lobe acceleration value is essential to determining required spring
force for any valve ensemble mass or cam lobe design.

Yes, that about makes sense. I'd just add that it's not the peak acceleration point or peak negative acceleration point where the valve loses contact. It's the point where the difference between the negative acceleration force exceeds the spring force. That's not necessarily at the same point at all rpms, and it depends on the spring forces.

I think I have a decent handle on how much spring I need when the valve is open and cam lobe has a negative acceleration. It's the seated load that's more difficult to me.

[ptuomov]

You allude to knowing what variable will trigger exhaust valve float in the current system. How do you know that?
Did you just keep turning up boost until it you lost power on the chassis dyno? Was it exhaust valve float, lifters or ignition?
What's the drive pressure at the higher boost level?

Anyway, figuring out your requirements from two directions and comparing results is better than only looking at one method, in my experience.

I don't really know anything.

I looked at the normally aspirated stock system, assumed turbos, changed the cams, reduced reciprocating weight by 20%, changed spring loads with custom springs, and figured out with my rules of thumb that the result should run well at 1 bar boost and 6700 rpm. It did run well at 1 bar boost adn 6700 rpm. Now, I am asking the question what should I change if I want the next engine with new rotating assembly to be good for 2 bar boost and 8000 rpm. I agree that thinking about it from multiple directions makes sense.

[Kevin Johnson]

Maybe you're jumping too far ahead here.

You're installing a new rotating assembly (I am assuming a different block etc.) that will be transmitting a new set of torsional inputs to the valve train via the timing chains.

Make sure that variable salad is similar to your current rotating assembly before invoking, "all other things being the same."

[digger]

To simplify the question further: ignoring the exhaust gas pressures and generally the gas pressure differential between exhaust port and cylinder, are the required valve seated loads proportional to the reciprocating component weight and proportional to the square of the engine speed?

its linear with speed

[gruntguru]

I looked at the normally aspirated stock system, assumed turbos, changed the cams, reduced reciprocating weight by 20%, changed spring loads with custom springs, and figured out with my rules of thumb that the result should run well at 1 bar boost and 6700 rpm. It did run well at 1 bar boost adn 6700 rpm. Now, I am asking the question what should I change if I want the next engine with new rotating assembly to be good for 2 bar boost and 8000 rpm. I agree that thinking about it from multiple directions makes sense.

I sympathise. Your approach so far makes a lot of sense and the results have confirmed the validity. Most builds arrive at spring selection by: experience, guesswork, trial and error or perhaps spintron testing.

A direct bucket-on-valve system has far less scope for component deflections and resonances and I agree with your proposal to extrapolate from the existing system to account for increased rpm and boost. It would be handy to be able to evaluate the resulting valve motion (detect valve bounce).

Is it possible to induce and detect valve bounce on the current engine? This knowledge would give you a more accurate starting point before extrapolating to the new operating point.

[digger]

To simplify the question further: ignoring the exhaust gas pressures and generally the gas pressure differential between exhaust port and cylinder, are the required valve seated loads proportional to the reciprocating component weight and proportional to the square of the engine speed?

its linear with speed

ignore this

[AC sports]

Stupid question.
For an equal weight and cam profile. Why do exhaust valves tend to bounce before inlets?

[David Redszus]

Stupid question.
For an equal weight and cam profile. Why do exhaust valves tend to bounce before inlets?

We are dealing with three variables; valve ensemble mass, valve acceleration and spring force.
If the above are equal, the inlet and exhaust valve will float and bonce at the same rpm.

But, when bounce does occur, the piston is closer to the exhaust valve than to the inlet valve.

[ptuomov]

Stupid question.
For an equal weight and cam profile. Why do exhaust valves tend to bounce before inlets?

My opinion (not a fact) is that in many engines the exhaust port and cylinder pressure have a larger differential on the exhaust side at EVC than on the intake side IVC. This may not be a large effect on a normally-aspirated even firing order engine with long-tube headers, but I believe it's a big effect in a twin turbo cross plane V8 engine with exhaust manifolds. And as was mentioned earlier, in pushrod engines the cylinder pressure can spring load the whole valvetrain at EVO.

[ptuomov]

Is it possible to induce and detect valve bounce on the current engine? This knowledge would give you a more accurate starting point before extrapolating to the new operating point.

Not with certainty. However, in similar engines that have exhaust drive pressure higher than the intake manifold pressure, when the exhaust valve bounces at EVC, the torque usually drops and the intake manifold pressure temporarily spikes as the exhaust gets pushed into the intake.

I don't think spintron is the answer here because I believe gas pressures are important. Therefore, one would have to instrument a running engine. Furthermore, because of the cross-plane V8 firing order, instrumenting just one cylinder would be enough, one would have to instrument at least two and possibly three most problematic cylinders.

[Kevin Johnson]

Stupid question.
For an equal weight and cam profile. Why do exhaust valves tend to bounce before inlets?

"Tend to"

Exhaust seats "tend to" be made of harder materials than intake seats. Assuming identical surface hardness of the valve faces the exhaust valves will rebound higher -- or initially -- simply as a material property.

Anvils are tested for hardness by dropping ball bearings on them and noting the rebound height percentage.

[MadBill]

I haven't seen any definitive data showing that exhausts do bounce higher or sooner than intakes. Since bounce can only occur after EVC and by then (40°+ ATDC) the piston will be out of reach of any still-connected ex. valve, gross signs of P-V contact can only be the result of float.

Float and bounce by the intakes leave more subtle clues and less immediate destruction, so judging only by the damage, it would be easy to conclude that exhausts are the weakest link due to sooner/higher bounce.

Also, typically, exhaust valves are lighter than same-technology intakes, plus commonly have less lift and more duration, both of which make the exhaust spring 's job easier than the intake's

[Kevin Johnson]

I haven't seen any definitive data showing that exhausts do bounce higher or sooner than intakes. '...

The Youngs Modulus will also decrease with temperature rise.

But then the loading per unit area on the seat might be greater.

But then different seat cuts are often used.

But then the length of the exhaust valve might be incrementally longer due to thermal expansion and affect the force applied by the spring(s).

Etcetera...

After running through arguments and counter-arguments in my head I decided that the principle of charity rules that I accept that the poster's presumably own- or third-party-observed data is legitimate and focus in on the qualifier "tend to." "Analysis-paralysis" should be avoided. Good scientific experimental method does dictate, however, that the researcher compile and present a list of reasons or confounds that could affect the validity of data or its interpretation.