A better explaination to exhaust scavenging

[josh817]

Alright so I finished this semester and during one of my courses we performed an experiment that I found useful for explaining scavenging.

 

This experiment was used to measure a shock wave as it traversed down a pipe. The pipe itself was split into two sections, one in which we pressurized with air, while the other side was at atmospheric pressure. A diaphragm was used between the two sections which ruptures at a certain delta pressure. This allows us to swap diaphragms to different materials or thicknesses to vary the rupture pressure and control the speed of the shock. For this experiment the diaphragm consisted of a plastic film which ruptured at ~160PSI if I remember correctly creating a shock traveling at around Mach 1.5. Similarly, hypersonic tests would be performed with a steel plate diaphragm that has a dimpled "X" on it to allow it to fracture at thousands of PSI causing a shock traveling at above Mach 15. Down the length of the pipe were 12 pressure transducers sampling at 100kHz. Seen below is a diagram of the shock tube and transducer locations. The right side of the tube is the side that is pressurized where as the left side is left alone.

 

 

As you can see, the left side has a cap on it with transducer P1. Two tests were performed, one with the cap installed and one without the cap, to study reflecting shock waves. For this discussion we will look at the test without the cap installed which would be a shock wave traversing down an open tube or if you use your imagination, an exhaust pipe (or intake runner).

 

If we plot the pressure data versus the time the experiment took place we would get this (noting this is data from transducers P6, P5, P2):

 

Obviously there is a spike but we can really see much. Notice it as steady at ~15PSI, that is atmospheric pressure. Also notice the pressure immediately after the spike, this is what we are interested in.

 

Reducing this down to focus on just the pressure deviation upon shock arrival we can see this:

Now we can see the shock as it passes each transducer. We know the distance between each transducer, therefore we can divide the distance by the time between the spikes and get a rough velocity of the shock. However, this still isn't addressing what happens after the shock. What happens to the pressure after it spikes and leaves the tube?

 

The pressure at P2, which is essentially the exit of the pipe, can be seen here:

 

See how the pressure spikes and then briefly dips below atmospheric pressure? Notice how it fluctuates slightly before steadying out. This lower pressure is what is responsible for scavenging. The effects become even more dramatic as you travel up the length of the pipe seen here at transducer 6:

The reverberation can be explained by the fact that the other end of the tube (the side that is pressurized) is closed. The momentum of the air rushing out of the pipe causes suction behind it, since there isn't any more gas to replace what has just left the tube. Once the shock has escaped the tube, the suction then draws in air where it flows back towards the ruptured diaphragm and fluctuates until it holds steady.

 

Our goal, to make scavenging successful, is to time this suction (vacuum, lack or pressure, whatever, your ass your face whats the difference) so that it reaches the consecutively firing cylinder as the valve is opening. You can see now that this would only really occur (the strongest) at one RPM where the timing of the suction matches the valve timing. This suction in conjunction with proper valve overlap will draw more aif-fuel mixture in and help evacuate the cylinder. Likewise on two strokes it helps evacuate the cylinder and draw excessive amounts of air-fuel mixture into the cylinder to the point where it flows into the exhaust pipe however due to the bee-hive design, it then forces this mixture back into the cylinder for combustion. This would most likely be utilizing the smaller pressure spike seen above, following after the suction essentially ramming mixture back in under slight pressure (albeit only 1-2PSI in this case).

 

Likewise this can also be used for the intake as the diaphragm on the shock tube doesn't have to have a pressurized side to rupture but rather a delta pressure. This means I can draw a vacuum on the left side of the tube while the right side remains closed at ambient pressure. When the pressure across the diaphragm reaches whatever the rupture pressure is, then it bursts. So, I can imagine lots of vacuum from an opening intake valve causing a pressure deviation (just like the pressure spike above) to traverse down a runner. Behind this vacuum spike there is going to be a rush of air causing a temporary pressure above atmospheric. The momentum of the air will cause it to slam into the now closed valve where it the pressure will spike again as the air is stacking up behind itself like train cars where it will then reflect off the valve and traverse back down the runner. The runner is open to the atmosphere so you will then have a case shown above where you have a pressure spike exhausting to the open air and reverberating back down towards the valve. This back and forth motion, as described above with the temporary vacuum with a lesser temporary pressurization continues inside the runner. This is what is happening when we discuss pressure cycles inside the runner and how the first cycle or wave is the strongest, the second becomes weaker, the third becomes even more weaker. You adjust the length of the runner to achieve whichever cycle you desire however to get cycles 1 and 2 timed with the intake valve would require very very long runners, usually too long for practical use. Like the exhaust, these events really only occur the strongest at a single RPM when the timing of the valve and cycles have aligned and the RPM you select to tune maximum effects for will also change the runner length. Basically, this means the runner dimensions (length and diameter) will be chosen by you dependent on the RPM you want maximum effects and the wave/cycle you wish to catch; Nascar guys typically go for the 3rd wave if I remember correctly.

 

A better visualization of this flowing air stacking up on itself like train cars, causing yet another increase in pressure behind the intake valve can be seen in the data collected for the experiment where we had the cap on the shock tube installed effectively creating an enclosed tube with a shock wave bouncing inside of it.

You can see how the initial shock arrives at each transducer and then spikes even higher also noting the order of the spikes which allows you to see the initial shock and the reflection as it traverses down the length of the pipe. Knowing these arrival times I can create a plot which allows me to select a location on the pipe and tell me when the shock or reflection will arrive with relatively....... good.... error. Kinda. The black lines represent the end caps of the tube with the "T" bars being error bars showing that the shock could be expected within that time frame. Note that the time scale is relative to the rupture time of the diaphragm, meaning that all the plots above were for a 2 second test however from the initial image posted we know that the diaphragm really only ruptured at ~1.2 seconds.

 

 

I will be looking at our other experiments to see if there is anything that can relate to what we are looking for when we design our motors.

 

Thanks for reading,

Josh

Edited May 5, 2014 by josh817

[johnc]

Great post and yes, the secondary pressure drops help extend the range of scavenging.

 

The general range for the primaries in a L6 header is 28 to 32" with some saying you can go as long as 36". That length is from the back of the valve to the last merge in the header collector or Y pipe. Longer is better if you want the extend the scavenging effect over a wider rpm range. But you will get a lessening of the peak due to reversion from the secondary pressure drops from other cylinders.