Valve Float Due to rapid Acceleration 3

[PFM]

All,

I have seen in print an article discussing valve float due to rapid engine acceleration at an RPM below that at which valve float occurred under a slower acceleration rate. I have seen evidence of this in engines on track that used low gearing and hard acceleration. I believe I have seen results of this in some drag applications as well. I can see why this could be true intuitively, the reason or calculating the why is beyond my rusty calc skills. I read the post about ignition errors due to rapid acceleration and it prompted this post. If possible I would like an equation I could plug in a known valve float RPM, an acceleration rate and the predicted float RPM. Hey may as well wish BIG.

Thanks in advance.

PFM

 

[ivymike]

inertia in the valve train vs valve spring tension is the sole cause of valve float... What have I missed?

Vibration of valvetrain components, entirely? As I mentioned above, jerk in the cam profile contributes to valvetrain vibration, and valvetrain vibration contributes to "float."

(If the kinematic acceleration of the cam hardware was all that you needed to consider, then valvetrain dynamics would be a much simpler subject)

 

[patprimmer]

Of course vibrations and harmonics, especially in the springs themselves, push rods if present, and timing chains if present. I think I already mentioned torsional vibrations in the crank and cam.

Maybe I am misunderstanding what you mean by jerk, maybe that means I am a jerk, but I understood jerk to mean a sudden movement when compared to the speed of normal movement, and I would expect that to be caused by torsional vibrations in the cam itself.

Maybe it also means what Rod was talking about where a lobe opening rate is deliberately designed to induce some float over the nose and thereby increasing lift, but that would happen at a specific rpm, irrespective of acceleration of the engine.



Regards
pat pprimmer@acay.com.au
eng-tips, by professional engineers for professional engineers
Please see FAQ731-376 for tips on how to make the best use of Eng-Tips Fora.

 

[ivymike]

nah, jerk is the time derivative of acceleration - the second derivative of velocity, and third of position.

 

[patprimmer]

So the jerk comes from vibrations in the valve train, and torsional vibrations are higher under conditions of acceleration.

Have I got it right this time.

Regards
pat pprimmer@acay.com.au
eng-tips, by professional engineers for professional engineers
Please see FAQ731-376 for tips on how to make the best use of Eng-Tips Fora.

 

[PFM]

Thanks all,

The input has been great but I still see no clear answer. Let me push the example a bit further. I hope my math is good. Again we accelerate the motor from 4000 RPM to 8000 RPM but this time in .0070 seconds. The valve and spring package is good to 8500 RPM. My math says at 4000 RPM one engine revolution takes .015 seconds, at 8000 RPM one revolution takes .0075 seconds. Now if the motor starts at one engine revolution in .015 seconds and ends at .0075 but the total elapsed time is less than the time required for one engine revolution at the end RPM at some point the instantaneous RPM MUST be greator than 1 revolution in .007059 seconds. All the input on valve float etc.. are great it all happens in an instant. Now the example is extream but...... I hope it serves the purpose.

Thanks again.

PFM

 

[ivymike]

PFM:

4000rpm = 1rev/0.015sec
8000rpm = 1rev/0.0075sec
4000rpm/0.007sec = 571428rpm/sec

Assume you have constant acceleration from 4000rpm to 8000rpm over a period of 0.0070 seconds. The engine will turn 252degrees in the time during which it is accelerating (0.007sec). The max instantaneous engine speed in that time is 8000rpm, which occurs at the end of the acceleration period.

to do the math yourself, for constant acceleration rate:
crank position = (initial position) + (initial speed)*(time) + 0.5*(acceleration rate)*(time)^2

for the example above,
252deg = 0deg + (24000deg/sec)(0.007sec) + 0.5*(3428571deg/sec/sec)(0.007sec)^2
252deg = 0 + 168deg + 84deg

Pat:
The jerk I was referring to is inherent to the shape of the cam profile, but its value depends on both the cam rotational speed and the cam rotational acceleration. The jerk contributes to valvetrain vibration, I believe, by altering the force-vs-time input to the system (changing the harmonic content of the excitation).

TVs might be higher during acceleration, but I've never run a transient analysis to look at torsional vibrations so I can't say.

 

[ivymike]

My math says at 4000 RPM one engine revolution takes .015 seconds, at 8000 RPM one revolution takes .0075 seconds. Now if the motor starts at one engine revolution in .015 seconds and ends at .0075 but the total elapsed time is less than the time required for one engine revolution at the end RPM at some point the instantaneous RPM MUST be greator than 1 revolution in .007059 seconds.

Just thought I'd add the following to my previous post. If we take your statement above, and translate it into more familiar terms, essentially what your math is saying is something like this: Driving down the road... At 60mph, one mile takes 1 minute. At 70mph, one mile takes 0.857 minutes. Now if the car starts at one mile in 1 minute and ends at 0.857 but the total elapsed time is less than the time required for the car to travel one mile at the end speed at some point the instantaneous speed must be greater than 1 mile in 0.800 minutes.

Does the above statement make the slightest bit of physical sense? Of course not- we know from experience that no matter how fast we accelerate, we're never going to go faster than the fastest speed we reach while accelerating, regardless of how quickly we're accelerating. A car that accelerates from 0-60 in 5 seconds doesn't magically go 90mph halfway in between, just because its speed increases by 60mph in less than the time it takes to drive a mile at 60mph!

 

[evelrod]

Well, Isaac. You managed to get me out to the shop and away from this keyboard. Now I have figured out two things: One---I haven't figured out what (or who) the "jerk" is so I'll assume it's me!
Two---Soooo, being the jerk ;-) here, I am sticking by my original assumptions (dangerous as that may be) in that I see no way varrying the acc rate can cause the lifter to float free of the cam lobe as long as the ultimate rpm where that float occurs naturally is not exceeded. I have some cam lobe profiles here that Ron Isky did for me back in the early 80's and they have some phenominal acc rates while keeping fairly mild overlap specs(DOHC application). These cams needed a bit stronger springs but no matter how we "flicked" the throttle, they never floated, indeed, valve float or seat bounce in that engine would have been disaster!

Rod

 

[PFM]

Mike,

Thanks for bringing out the 2X4, sometimes when my record gets stuck it needs a tap back to reality. The car deal got through my fog and made the point.

Rod, As for your cam and not floating the valves with a flick of the throttle well that is a whole new thread too, what if the valve floated just after the opening ramp and was not near the piston when it floated?

Ok, Thank you all for the help with this thread.

PFM

 

[ivymike]

what if the valve floated just after the opening ramp and was not near the piston when it floated?
Then you'd have a REALLY odd cam profile.

Rod, I certainly can't come up with an actual situation from memmory where the camshaft angular acceleration would have had a big effect on valvetrain performance. I can't rule out that such a situation might occur given the right confluence of circumstances, though, because I know that the jerk of the cam profile can be very important, and I know that the jerk is related to both cam velocity and cam angular acceleration. Cam jerk won't ever cause separation in a kinematic situation - only cam acceleration can do it there - but cam jerk can definitely cause severe vibrations in a dynamic situation, where the various valvetrain components can vibrate significantly. I haven't done much with OHC engines, but I suspect they're much less sensitive to this than pushrod engines. All that said, I'm not certain that camshaft acceleration (within reason) will have any significant effect on cam jerk.

 

[Flatjwl]

It is simply not possible to accelerate the valve more rapidly on the way up to maximum RPM, than it is accelerated while operating at maximum RPM.

There is a useful form of valve float which I refer to as valve "loft" and it can improve performnce when used in a controlled manner.

Regards,

John Lawson

 

[PFM]

John,


"It is simply not possible to accelerate the valve more rapidly on the way up to maximum RPM, than it is accelerated while operating at maximum RPM."


You state that with such conviction. Though IvyMike has given me an example I think I can live with I still have some issues I cannot resolve.

Lets move to a super extream example, 4000 RPM to 8000 RPM in .00001 seconds. Now if we apply Mike's formula to this all looks fine but can it be?

The formula would give a result that the crank did this in like 1 degree of engine rotation, yet it does not have an effect on valve opening, broken rods, pistons etc..

This question has been proposed to more than one engineer (not on this forum) and they were not able to prove or disprove the concept. That is why I tried it here.

Thanks again to all.

PFM

 

[evelrod]

Yes, Isaac. The DOHC configuration is much less suseptable to "bad" harmonics than a conventional pushrod engine, IMO. That given, most of what I have learned on the OHC setups is readily transferable to other concepts, I assume. I did witness a strobe test on a BBC in the earlly 80's , a single plane, 4xx cu in sprint car engine that I thought impressive. It had BIG pushrods (3/8"?) and under the strobe they looked "bent"---big wobble at speed! From that I would assume the spring pack was doing a dance, too. That given, I can see where all sorts of tappett cam acc ramp interface problems could exist. I can't argue the "jerk" problem as I simply cannot grasp the concept aside the usual timing chain "flux" of normal opperation. In setting up the ign on my last 1600 I only got a ~1/2 degree pk to pk on the dizzy vs. the crankfire. The dizzy is driven from a jackshaft, no lobes and the timing chain is a 35 Morse, pretty long but with a tensioner. Can I assume that 1/2 degree is due to the jackshaft being driven on the "up" side of the cam chain? Chain route is crank, ex cam, in cam, tensioner, jackshaft, crank, etc. It has never been a problem so I have never even looked at it.
As to pushrod engines, you already know that is pretty much limited to the mini--- all the head work is now done by APT---it's just too late for me to start at the bottom of the curve again. $1500 will get a top notch head and I could waste that much just getting the ports to flow the same.

Good morning, John Lawson. As to the "loft" thing---as I posted above, I have some small degree of experience in grinding a cam lobe profile that enhanced that little phenom, albeit in the dark ages.
One thing I did learn---it is not just the acc ramp/opening side of the profile that causes the "good float", the closing/decel ramp was also a determining factor. As I said, this worked well on a 5 hp (to start with, anyway) Briggs-Stratton side valve engine and it did not work at all on a couple of automotive engines where I got the idea in the first place. This was all 1960's stuff and is totally out of whack today. I would never think of doing something like that on one of my 10,000 rpm engines now.

Rod

 

[ivymike]

It is simply not possible to accelerate the valve more rapidly on the way up to maximum RPM, than it is accelerated while operating at maximum RPM. Considering only kinematics, you are correct.

 

[strokersix]

Now wait a minute. I've been reading this thread and resisting comment until now.

In a simplified (but technically correct in my opinion) linear model crankshaft motion will be superimposed on valve motion. This means that crankshaft acceleration will be added to valve acceleration.

Whether this is significant or overshadowed by other effects is open to question. I suspect that other effects such as those discussed above in previous posts are dominant.

 

[ivymike]

Maybe I don't get what you're saying. The crankshaft goes 'round in circles, while the valves go up and down. How is the crank motion superimposed on the valve motion?

Try this - draw an acceleration vs crk angle diagram for a single valve at a single engine speed. Note that for large portions of the cycle, valve accn = 0, regardless of what the crankshaft is doing. Also note that for zero crank acceleration, there is significant valve acceleration.

The valve acceleration diagram will look like a distorted letter M, with a positive hump followed by a negative-amplitude valley followed by another positive hump. The first hump is the opening flank (and ramp), the valley is the cam nose, and the second hump is the closing flank (and ramp). The amplitudes of the humps and valleys are determined by the shape of the cam, and are linearly proportional to cam velocity (prop to crk vel). Thus we can see that kinematic valve acceleration is proportional to (average kinematic) crankshaft velocity. A little calculus will show that the first time-derivative of the cam acceleration profile (cam jerk) is proportional to the first time-derivative of the crankshaft velocity (crank accel).

In the kinematic model above, separation only occurs when the amplitude of the negative valley @ the cam nose is greater than the max acceleration amplitude that the springs can impart to the valvetrain hardware. In other words, if |the spring force at maximum deflection divided by the effective mass of the components| is less than |the required nose acceleration|, separation will occur.

In a dynamic system, all of the valvetrain components are flexible and have mass, and all can vibrate individually as well as in a group. If you look at a dynamic valve acceleration trace, there will be a wavy pattern superposed upon the kinematic "M" pattern. The valves will open a bit later, and close a bit earlier (or later) than expected. Peak acceleration is no longer necessarily at the same point in time as it was in the kinematic model. The amplitude of the max acceleration is no longer directly proportional to crankshaft velocity either because the vibrations will increase and decrease as various vibratory modes manifest within the valvetrain at different engine speeds (although the underlying kinematic acceleration still is prop to crk vel).

These vibrations are driven by the force input to the valvetrain from the cam lobe, which is time varying (of course). The frequency content of the excitation will influence the amplitude of the response at any given engine speed. The frequency content of the excitation is influenced to some extent by the camshaft angular acceleration, which is why I don't want to rule out engine accel as a possible contributor to valve "float."

 

[ivymike]

A little calculus will show that the first time-derivative of the cam acceleration profile (cam jerk) is proportional to the first time-derivative of the crankshaft velocity (crank accel).

I don't think my statement above was quite correct - crank acceleration will influence cam jerk, but they're probably not proportional. (I haven't actually written out any equations and solved)

 

[strokersix]

ivymike,

I think you do understand what I'm saying.

I'll restate for the benefit of those who perhaps don't:

The valve position, velocity, acceleration, jerk, etc. are tied directly to the crankshaft. Crankshaft motion is superimposed on valve motion. If the crankshaft is going faster then the valves are too. If the crankshaft accelerates then the valves see higher acceleration and lower deceleration in an additive manner. Likewise, crankshaft torsional vibrations are also superimposed on valvetrain motion.

Obviously there is a lot more going on with regard to vibrations, deflections, etc. but I believe my statements are technically correct.

 

[SUPERKYLE]

All,
I am not a cam designer, nor a rocket scientist but I did stay at a Holiday Inn Express last night. No actually I have been involved with valvetrain testing using the Spintron test equipment.
The acceleration in question is the engines acceleration rate which is variable even at W.O.T. based on load, gear ratio and power curve shape. If I remember right our T/A engines accelerated around 2500 rpm/sec in 2nd gear 1200 rpm/sec in 3rd 600 rpm/sec in 4th and 300 rpm/sec in 5th.
My experiences tell me the following;
If the valvetrain has a period of instability before maximum rpm, the engine can pass through it more easily the faster the engine is accelerating.
Loft is your friend. Its the landing that's the problem.
Bounce is what happens when the ski jumper lands on the flat. (BAD)
Piston to valve issues occur only at TDC overlap. The cam position at overlap has the exhaust just closing and the intake just opening. So valve bounce can only affect exhaust piston to valve clearance.

Later,
Kyle

 

[1Bluegrass]

I totaly agree with Kyle. All I have studied and worked on about valve control in normal street engines and race engines under 10k rpm involve sufficient valve spring pressures for the rate of lift going over the nose of the cam vs useful rpm power band of (solid lifter cams).
The valve set-down can be a seperate issue also involving the flank set-down rate causing valve seat bounce.
The spring setup still has effects on this under different conditions.
One thing I have noticed on many different head designs is the possible tendency of the head material and design in the area of the valve seats contributing to the bounce issue with it's own 'ringing' reaction after inital hit by the valve head.