Compression Leak or Not? Math for the win!

Dmitriy Analyst Toronto, Ontario Posted   Latest   Edited  

A while ago I prepared an article 'Practical Compression Calculations', but posted it as a resource as I did not have a good case study to demonstrate it on. Until now.


Recently, in another thread, Jonathan referred to one of his case studies where two in-cylinder captures are performed simultaneously in order to diagnose a turbocharged gasoline engine that misfires at high RPM: 


Undoubtedly, it is a fascinating study, so let's read it up until page 6 and focus on the zoomed waveform:


I loaded it into a graphics editor and marked up a few points:

  • A : first compression peak of the blue trace;
  • E : second compression peak of the blue trace;
  • C : compression peak of the red trace, confirmed to be 180 deg before point E;
  • D : clearly visible intake valve close for the blue trace;
  • B : should be intake valve close for the red trace as time from B to C is the same as time from D to E.

The pressure at point C is much lower than at point E, and to many diagnosticians this would look like a compression leak... but let's see what the math says.

From the Practical Compression Calculations post we know that if the intake valve for the red trace did close at point B and the compression stroke continued on normally, the ratio of pressures (in psia) between points C and B would be close to the ratio of pressures (in psia) between points E and D. Let's calculate those ratios.

It would be easier to do this if I had the PsData file, but I can work with pixels too (not as precise or convenient, though). This is how I identify the pressures using a spreadsheet. I record positions of the blue and red scales, positions of the points A-E, use proportions to identify pressures in psig, then add 14.7 to get psia values:


Aaand... the ratios turn out to be similar; the are no leaks during the compression process if it starts at point B, so point B should become the focus of further analysis. How come the pressure at point B is lower than at point D? Looking at point D we see that the pressure of the blue trace has stabilized just before point D, presumably at the intake manifold pressure. But this is something that never happened for the red trace before point B, with the pressure creeping up relatively slowly after the intake pull.

The waveform suggests there is a cylinder intake restriction, but what kind -- whether it is the intake valve opened only half-way or something in the intake port -- I can't claim to know. If you have ideas on how to distinguish between these possibilities, please share. In Jonathan's case study the disassembly and visual inspection revealed the culprit.

Joel Diagnostician
Soquel, California
Joel Default

very nice, I will keep reading this to absorb more . thanks for sharing

+1 Default Ð Bounty Awarded
Bill Technician
Rosetown, Saskatchewan
Bill Default

I just skimmed through this, so maybe reponding at this point is a bit premature. If the intake was only partially opening you would expect to see it close early. Thats what I would be looking for. Dmitiry, something I have always been very interested in is the amount of pressure in the cylinder at intake valve closing under WOT max cylinder pressure RPM. With a WPS there is a certain amount of…

+1 Default Ð Bounty Awarded
Dmitriy Analyst
Toronto, Ontario
Dmitriy Default

Thanks, Bill, what you are saying about the intake valve makes perfect sense, but I just don't know all possible designs of actuating the valves well enough to exclude possibility of such failure mode 100%. So I'll gladly let people who have seen it all take the lead. You are right that the pressure around the IVC moment may not be stable and all calculations would be quite imprecise as a…

0 Default Ð Bounty Awarded
Bill Technician
Rosetown, Saskatchewan
Bill Default

Dmitriy, Any data I have looks inaccurate. At one point I took in-cylinder readings with CKP to see what type of WPS lag I had during the most efficient pumping speeds. I disregarded it as inaccurate. Not sure the equipment I have is sufficient for your calculations.

0 Default Ð Bounty Awarded