All the discussion of idle and delta tables was an aside that I shouldn't have introduced. The point above is accurate for a homogenous fuel mixture. Flame temperature drops, but so does flame speed leading to slower combustion and higher mean cycle temperatures. In an engine with a stratified fuel charge (like diesels, Hondas with VTEC-E, and the 13B-MSP) combustion takes only slightly longer.

https://www.rx8club.com/attachment.p...1&d=1305237724
Kawahara, et al, Proceedings of the Combustion Institute, 2007

Again, this is a tangential issue, but if you use negative split in the main timing tables and don't change the idle tables, it looks like this:
https://www.rx8club.com/attachment.p...1&d=1305240170

I thought the idle felt weird that way, but I guess it is subjective.

That's an interesting idea. Negative split might greatly increase post oxidation, and create a "false" lean condition. If true, it would make sense for Mazda to use that trait to help out on 2-speed idle emissions tests. Maybe we are actually idling at .80. :)

Max fuel efficiency being at 16.x makes sense for aviation, when you are talking about what fuel consumption is needed to maintain a given air speed.
Think the problem domain is different for cars with fixed transmission ratios. Operating in lean burn gives another way to vary output so that it is closest to the minimum required to maintain vehicle speed at a fixed engine speed. Anyhow, I know Honda's operate at AFRs in the 20s.

For anyone wanting to try this, I think the best solution would be to get a wideband controller with programmable outputs, like the Innovate LC-1, and connect one to the ecu in place of the stock one.

However, I've just confirmed that you can run any AFR you choose by eliminating closed loop and de-calibrating the MAF scale changing the VE tables. A precisely jetted carburetor is what it'll be like then.

I measured while stuck in a traffic jam on the Tappan Zee bridge in New York state where I didn't move for half an hour.

subscribed to read later

For the time I was willing to try it, cars runs nicely at indicated 19:1.
Realized I have nearly no clue how the closed loop works, there are many different schemes it seems, none well-documented that I can find.
May have to dive back into IDA Pro, if I can't suss out the logic empirically.

So, you are measuring fuel consumption as a function of time?
So, go ahead and do that for the rest of your driving and see what you come up with. You may be surprised at how good your consumption rate is at idle.

Otherwise, turn the engine off when you are just sitting there.

The main differences between autos and planes I'd say is the aircraft: (1) run for hours with the same engine output (2) can choose their speed (3) have bad things happen if the engine stops. With the Mooney (and its variable-pitch propeller (think CVT)), I could cruise at 120 mph @5 gallons/hour, 150 mph @8 gph, 180 @14 gph or anything in between. With a C172 (fixed-pitch (think one gear)) I could do something quite similar. The difference is, that for a fixed airspeed, the ability to adjust the propeller gives much more flexibility in achieving the needed power output from the engine.

A car cruising at a constant 70 mph presents a more difficult problem in that the power has to be adjusted quite frequently. The well-recognized most efficient solution is to freeze the throttle position (~constant power output) and let the speed vary as the road rises and falls. The advantage with cars is that they have much more sophisticated engine management controls than the entirely manual controls on most aircraft.

The latest development in car engine management is the trend to replace the direct link between the throttle pedal and throttle plate, whether mechanical or electronic, with making the throttle pedal a power-command servo. In other words, when the driver pushes the pedal, the request goes to the ECU which then adjusts engine parameters (including possibly, but not necessarily, the throttle plate) to achieve the desired power. The RX8 of course is not this sophisticated, so adjusting afr as a way to vary power output is I suspect not easy to fine tune. As such, what I'm hoping for is a map for daily driving to get best mpg's and a different map (MM) for best power, loading one or the other as needed. The Cobb "Economy Tune" hints at that solution, though opinions vary about its effectiveness.

I don't think there is much need for two maps. The ecu already has so much flexibility. Optimize low-load and closed loop for fuel economy, and high-load open loop for power. On a track you will almost never use the former, and if you drive conservatively, the latter.

Right now, I target 1.05 lambda in closed loop. With the factory wideband, the ecu can maintain this.

Lean-burn works by reducing pumping/throttling losses at very low loads. If you need 20hp to maintain cruising speed, and 5hp is wasted creating vacuum pull air in, overall efficiency increases even though you get less energy per unit of fuel at 20:1 than at a richer mix.

Rotaries have great flexibility in intake design, as good as the RX8s intake is for power, it could probably be better at low loads. I'll just go ahead and trademark "Portronic" right now.

Correct.

Wonderful theory, but you are going about it incorrectly.

The best lambda for economy - established through hundreds of hours of testing - is between .92 and .95.
Nominal torque is more important that fuel/time.

The way you establish your "lean-burn" is through timing split.

That said, the heating that happens in the combustion chamber - while helpful for economy and emissions - is dangerous with regards to flame front speed when the charge homogeneity and load suddenly increase (and lambda decreases). When the throttle is suddenly open and everything changes, that heat becomes immediate detonation.

That is too bad. If negative split is not useful for maintaining lean burn, can you explain how Mazda did it with the lean burn RX4 that they released in 1980, or how they are doing it with their prototype Renesis?

http://www.mechanicalengineeringblog...enesis-engine/

That would be useful.

I have run my motor at an indicated 19:1+. Negative split timing did not change the indicated AFR per se.

Re-read post #48.

I guess you don't know anything about this topic, thanks for trying.

Perhaps you might do well to get out of your parent's basement.

Also, as an advisory to anyone following along, cruising lean produces more oxides of nitrogen (though rotaries are known for producing lower levels), and the reduced CO levels will reduce the efficacy of the rhodium catalyst.

However, cruising rich will increase CO and HC output, while also greatly heating and potentially causing substrate meltdown. Do not attempt to cruise rich unless you are using a midpipe.

Some people seem to have interpreted this as a fuel economy thread, if you're reading just for that reason I don't want anyone taking any bad advice.

Ouch! Actually, they don't have a basement, but I'll be driving to a party in Echo Park tonight. Should be on the freeway for at least 100 miles, I'll post my mileage either way. What are you up to this evening? :)

The CAT was expected to operate in an .90 lambda environment for 100k+ miles. It really isn't part of the equation.

The point I was making was that the actual combustion AFR is not the observed AFR from your example because of the effect of the negative split. You don't have enough data to post a useful observation from your example.

I have BTA (Beer Time Attack) with the NASA peeps and some of the RTA competitors tonight.

and the crowd goes "ooooo and aaaaaah" :lol:

lol

Well, he asked!

Found that LTFT corrected my AFRs after about 10 mins as noted by Oltman .

So - no dice

Got 20.6mpg using the lean limit for CL on a 90% hwy trip. Essentially the same as I've gotten on similar trips in the past. I blame E10 and my habit of going 85-90. Used stock timing.

Seeing no gains, I decided to try targetting .95-92 in CL above .50 load. Made me realize that stoich or leaner really sucks with just about any kind of acceleration at all. No results on that yet, but I don't really care. Yes, you did tell me so, MazdaManiac.

I've read that Honda Insights run stoich in CL even at WOT. Sounds punishing. I've realized that stoich is good if you used lots of Aquanet in the 80s and you feel really guilty about what you did to the ozone layer. Otherwise, it sucks.

I still haven't satisfied my intellectual curiosity about lean burn, but I'm not sure I'm willing to blow two bills on a programmable wideband right now. I guess I'll just read this 315 page report on the NASA SCORE project for the time being.

http://protonet.org/doc/NASA_1994029725.pdf

It occurs to me that Mazda might be working on lean burn just for emissions reasons. All emissions drop when you go really lean, add a two-way cat and a NOx trap and you make the EPA happy. Might be a needed change to keep rotaries viable regardless of whether lean burn gives rotaries the same fuel economy gains as piston engines.

Think you will find (as I did) that the actual AFRs are being brought back to 14.7 by the LTFT .

As has been pointed out, I have no idea about "actual" AFRs. :)

However, my ecu (2005 federal) limits closed loop to a mid-15 reading from the WB O2 sensor. I can target anything up to that in closed loop targets A, and it works apparently indefinitely, for many hours at least. Targeting higher only works for a split second, although I did once have it work for longer. I believe that happened because it was just above the exit load for closed loop, and just below the next open loop load point. Programmers call this a fencepost error.

I don't mess with the other closed loop table anymore, I don't think they are fuel targets, but rather set feedback gain for the reading from the wbo2.

Or are you comparing readings from the stock wbo2 and an aftermarket one?