Water hammer principles in gaseous flow 1

[BillyShope]

I was an engineer at Chrysler in the fifties. (Yes, I'm that old.) At that time, it was quite easy to talk with other engineers about dynamic compressibility effects in gaseous flow. In other words, we realized that the sonic wave nonsense in Phillips' book simply could never explain the dyno numbers we were seeing. So, we soon realized that all we learned in school about water hammer had to be applied to that OTHER fluid: A gas! When we started applying the water hammer equations to the gas in a tuned intake manifold, the pressures all began to make sense!

One easily recognized example was the first club car assembled by the Ramchargers. The intake manifold was taken right from a dyno room. The eight exhaust pipes ended with cones for the obvious purpose of reducing energy loss during reverse flow.

So, the obvious question is: Why are the water hammer effects apparently ignored today when it comes to the design of intake manifolds?

 

[VEBill]

Perhaps superseded [<- edit, spelling] by Computational Fluid Dynamics (CFD) ?

 

[RodRico]

More properly, incorporated into Computational Fluid Dynamics?

 

[BillyShope]

Unfortunately, CFD is of no assistance. We've passed the problem on to the programmers and dynamic fluid flow compressibility programs are often difficult to write. In this case, we're dealing with the existence of what are often called harmonics. For instance, Offy engine owners in the fifties usually tuned injector stacks to the fourth water hammer harmonic. Chrysler later used those increased benefits when tuning to the third harmonic.

 

[RodRico]

BillyShope,

CFD software is routinely used to model intakes, and even the most rudimentary analysis incorporates pressure waves due to dynamic conditions. There are many videos of such simulations on YouTube. The one at

https://youtu.be/X3ecq8MNfk0

is particularly nice because it illustrates how it's done using my toolset (Solidworks) with Excel generated tables controlling boundary conditions over time.

Rod

 

[BillyShope]

Rod, that which needs to be considered are the pressure fronts which traverse the intake runner while the intake valve is closed. That's pressure FRONTS and not pressure WAVES. The number of water hammer cycles ("water" being an unfortunate misnomer) determines the "harmonic" under consideration. If roughly three such cycles pass before the valve is again opened, we consider the manifold to be tuned to the third harmonic. These pressure fronts are not to be confused with any sonic "waves." The pressure front might define a difference of 3 or 4 psi; a sonic wave involves a negligible pressure difference.

 

[SwinnyGG]

You're describing the same thing. A pressure 'front' is a wave. This behavior is well understood and is not neglected.

 

[gruntguru]

As a matter of fact, the phenomenon does not even require 3D CFD. 1D engine modelling software handles wave/front/inertia tuning more than adequately.

je suis charlie

 

[RodRico]

I know how to use Nyquist theory to analyze the 4th harmonic of waves, but I’m unclear how to analyze the 4th harmonic of a “front.” In any case, you can watch numerous videos of CFD applied to the actual water hammer problem on YouTube similar to the one at

https://youtu.be/sb_LFUkh4qA.

Notice it actually shows the pressure waveform, and it definately has a lot of harmonics because it’s as much a square wave as a sine wave.

 

[SomptingGuy]

The effects are captured in 1D CFD codes well enough for them to be used to design and calibrate things like this: F1 intake trumpet actuators. Length is varied with engine speed so that wave interactions are always beneficial.

Banned shortly after it was perfected. Then made obsolete when turbochargers were reintroduced.

Steve

 

[BillyShope]

Rod, I can understand your confusion. Back in the fifties, I objected when others at Chrysler used words like "harmonics" when describing tuned manifolds. Water hammer fundamentals are confusing enough without the idea that we have to call on music majors.

 

[BillyShope]

But, it must be emphasized that a "wave" most certainly cannot be assumed to be the same as a "front." When we speak of a wave, we normally would picture a pressure that is varying in magnitude as a function of location along the runner. With water hammer, the pressure is essentially constant between regions which I choose to call "fronts."

 

[BillyShope]

I assumed this was understood, but I am assuming that the valve is closed.

 

[SwinnyGG]

With water hammer, the pressure is essentially constant between regions which I choose to call "fronts."

No, it isn't. It just appears that way when you use a first-order approximation (i.e. a simple wave which does not include the large number of nested harmonics).

The ability for simulation of actual particle paths and pressures, even in the 1D space as noted by gruntguru and SomptingGuy, is very high in the current engineering environment, and has been for quite some time.

 

[RodRico]

I’m not a “music major” but an engineer. The water hammer effect is much like ringing in a physical system or impedance mismatch in an electrical system. In this case, it results from interruption in flow that yields pressure as fluid momentum encounters a hard stop. The fluid then seeks the lower pressure region back toward the source. There, it encouters resistance at the source and propagates back again. The process continues until equilibrium is found. Observing pressure at any point in the path yields a waveform of rising and falling pressure over time that reflects the sum of all pressure “fronts” moving back and forth through the pipe, the length and diameter of the pipe combined with fluid properties defining the waveform shape. This pressure waveform can be decomposed to sinusoidal frequency components via fourier analysis.

 

[BrianPetersen]

The engine simulation program that I've fiddled with - the free version of Lotus Engineering! - can provide graphs of the pressure at any point in the system as a function of time. That closed-valve resonance inside the intake runner, against the back of the closed valve, is most certainly evident in the simulation results.

I've also never found it particularly useful to pay attention to the closed-valve effects. What may be the third reflection at 10,000 rpm and producing a positive pressure in the port during valve overlap may be the fourth reflection at 7,500 rpm, and halfway between the two (at around 8,800 rpm) it's a negative pressure and acting against you, and then it's acting against you again a bit beyond 11,000 rpm. I need a wider powerband than that. Peaks and valleys through the normal operating range net out to zero - or worse, if one of the "bad" valleys is bad enough to cause a driveability problem.

The "ramming" effect near the end of the intake stroke is far, far more important, and is useful over a wider operating range. The wave effects during the single intake stroke are also useful over a wider operating range than the closed-valve effects are.

 

[BillyShope]

OK, I wasn't planning on doing this, but I'm going to have to present things as I would back in the fifties and then we'll try and work out a translation.

For calculations, I used a K&E Log-Log Duplex Decitrig slide rule. Many years later, I'd get a self assembly electronic pocket calculator from Great Britain. Chrysler's Engineering Center in Highland Park had an early IBM tube type computer (I've forgotten the IBM designation). In some ways, the computer was pretty crude. I mispunched a card and the computer was down for a couple of days. It was often more convenient to use a Friden to make vehicle performance calculations.

Anyway, here's how we visualized the workings of a tuned intake manifold. Valve closures and openings are assumed instantaneous. At the instant of valve closure, the intake runner is at zero pressure and the flow is toward the valve. The air (actually, a F/A mixture) impinging on the closed valve immediately rises in pressure to a value equal to ρVc or, in other words, the product of the specific density, the flow velocity, and the speed of sound.

So, this very small volume of high pressure air immediately begins to expand toward the open end of the runner, or, in other words, away from the closed valve. There exists, then, a zero thickness boundary (that which I call a front) moving away from the closed valve. Behind the front, the air is stagnant and at a high pressure. Ahead of the front, the pressure is zero and the air is flowing toward the closed valve. The front travels at the speed of sound.

When the front reaches the inlet of the runner, the runner is filled with high pressure stagnant air. At this point, the phenomenon of reverse flow begins. Air begins to flow in the runner, but, instead of flowing toward the valve, it flows from the runner back into the atmosphere (or plenum, if the engine is so equipped). So, another zero thickness front exists, with the air on the inlet end of the runner, flowing back (reverse flow), at zero pressure, into the atmosphere.

When this newly described front reaches the closed valve, the pressure of the air near that valve head acts as you would expect. With the air now flowing away from the closed valve, the pressure drops by that same ρVc amount. When this new front reaches the runner inlet, it should not be surprising that the initial flow conditions reappear.

It is apparent, then, that there exists an optimum number of cycles between valve closure and valve opening. This is when engineers start talking about “third harmonic” tuning and “fourth harmonic” Obviously, we would choose a minimum number of wave traverses. Unfortunately the manifold becomes more difficult to stuff under the hood as we reduce traverses. Jaguar experimented with first harmonic tuning on the dyno and got volumetric efficiency numbers above 100%, but the third harmonic seems to be a good compromise. Of course, that nasty Second Law places some limitations on all of this.

There were equations tossed about at Chrysler, but I have deliberately avoided them here. Well, I'll throw in one: NL = 84000. However, this is only somewhat valid if you're using very “tame” cam timing. The article about “Ramming The Rat” in that highly technical magazine “Hot Rod” (available online) is worth the reading. Yeah, I'm the old guy in the pictures.

 

[BrianPetersen]

You can be certain that the phenomenon you describe is most certainly accounted for, even in a basic 1-D simulation ... and without the simplifying assumption of instantaneous valve events.

I didn't find it useful to tune for that phenomenon, because of the peaks and valleys through the operating RPM range that my applications require.

 

[BillyShope]

"I've also never found it particularly useful to pay attention to the closed-valve effects." This would indicate that communication is impossible. As the old saying goes: If I were to agree with you, we'd both be wrong.

 

[BrianPetersen]

I'm interested in your explanation of what happens with that effect at an "off-design" RPM, nevertheless. If you calculate the RPM at which you have (let's say) a positive influence at the 4th reflection, and the RPM at which you have a positive influence at the 3rd reflection, what happens at the RPM halfway between those two, and what would you suggest to do in order to deal with that?

I know what the simulation says ...