Flame Velocity vs. Ignition Timing

[carwhisperer]

I am trying to do some calculations to estimate/replicate the timing table for an LS1. Here are my assumptions. The spark plug is offset approximately 1/3 of the way across a 99mm bore, so max flame travel distance is 66mm (I got this by estimating from a picture. Stock timing for 600rpm is 11 BTDC, and at 6400rpm it is 26. This is from a table I found online and is at .72g/cyl.

So, assuming flame termination is at 15 degrees ATDC I determined the flame velocity to be about 9.1 m/s at 600 rpm. Then I used a graph from Pulkabrek's ICE book and got a slope of about .02375 m/s/rpm to estimate how much flame velocity increases from 600 to 6400. This gave me a flame front velocity of 146.75 m/s. This puts timing at only 2 degrees BTDC, obviously not right and far off of the table's 26. I tried another method from the book, which is to scale the ratio of the rpm's with a factor of 0.85. This gives a velocity of 82.9 m/s and flame travel time of .000796s and leads to a timing of 15 BTDC, which is better but still way off of 26, which seemed like a low number to begin with.

What am I doing wrong? Am I way off on the flame travel distance of 66mm because of the pent roof combustion chamber? I would assume that the table would have been generated more or less empirically, but it seems like there should be a way to get a lot closer than I got.

 

[gruntguru]

Your assumption of flame termination 15* ATDC is one issue. Perhaps around 80% of mixture mass will be burned by peak cylinder pressure which is often assumed to be optimally located at 15* ATDC. Google images for "Mass Fraction Burned". Link
Difficult to infer the position or velocity of the flame front from MFB though. The last mixture to burn is very slow due to chamber wall quenching. Perhaps the velocity for the first 80% burned is a good starting point.

je suis charlie

 

[Overrun]

Keep in mind that your original flame kernel has a very small front and burns in three dimensions. It will burn the bulk fuel charge much slower than a later large flame front. On the other hand there is also likely squish and/or tumble in the fuel charge that changes the flame travel from conduction to convection, i.e. maybe a 10X to 15X increase in flame travel rate.

 

[carwhisperer]

Ok that makes sense. What sort of algorithm can be used to arrive at more realistic timing numbers?

 

[BrianPetersen]

Lots of dyno testing.

 

[carwhisperer]

There's got to be some model that can get you in the ballpark, right?

 

[Tmoose]

Highest pressure ( at your estimated 15 degrees ) does not mean combustion is over at that point, so I wonder if the flame ain't still travellin' .

I think combustion in a gasoline engine requires a few "stages."
- Spark
- ionization
- initial slow kernel growth sufficient to create any pressure at all
- usefully fast flame "propagation" if swirl, tumble and turbulence to wrinkle and fold the flame front sufficiently to compensate for the turtle-like basic "flame speed" of an air gasoline air vapor. (Only a foot or so per second, according to Kevin Cameron).

 

[BrianPetersen]

I am sure that the engineers who are in this business have an extremely complex computational fluid dynamics model that includes combustion simulation which might let them get close to an estimate of what the ignition timing should be under various conditions.

The rest of us make do with a starting point based on previous experience with this type of engine, guess on the slightly late side to avoid melting a piston before even getting started, and then do lots of dyno testing to refine it.

The computer simulation will not get you out of doing the dyno testing.

There is absolutely no way you will ever just plug in some simple rule of thumb formula and expect results that are anywhere close to being accurate. It doesn't work that way. There are too many variables and uncertainties. Squish clearance. Squish band area. Initial tumble or swirl in the incoming charge. Fuel distribution in the incoming charge. Spark plug location. Operating temperature of all of the surfaces in the chamber.

 

[Panther140]

The general rule of thumb here is that there are no rules of thumb in your calculations. They use pressure transducers in the cylinder head and combustion imaging cameras. Hard data on the crank angle domain. That's actually what I'm doing right now.

Educated guessing will get you as close as the calculations will, because you're going to guess on a lot of your variables.