subaru Impreza: lean burn and valve lift experiment
 

I have been self tuning my impreza for power for the past couple years and learned a lot in the process. But the worst gripe I have about my car is mileage. I made some tuning posts before but I never had this much tuning 'power' as I do now. i can change anything that is ECU controlled including but not limited to EGR, lean burn (AFR), timing and my favorite valve lift which subaru calls i-AVLS (exclusive to their NA SOHC powerplants).
http://i.snag.gy/GfLKV.jpg

this is same as Honda's VTEC but restricted to intake side only. So it opens or nearly closes one set of the intake valves.

High-lift
http://drive2.subaru.com/Spring07/avls_high.jpg
The rocker arms of each cylinder’s two intake valves are locked together, and the high-profile camshaft lobe operates them both. The two wide valve openings reduce resistance to intake air, helping to improve horsepower.

Low-lift
http://drive2.subaru.com/Spring07/avls_low.jpg
Each intake valve is operated via rocker arm by its own camshaft lobe – one lobe with a high profile and the other with a lower profile. The resultant staggered valve openings increase the speed of the air entering the combustion chamber, which increases torque.

Stock valve lift is set to a questionable 1500 RPM!! This seems to only help with idle emission and quality. However I raised this up to around 2900 RPM which would equate to about 75 MPH on the highway (which is the fastest I go on the streets). Now I have to open the throttle more to make the same power or hold the same speed on the highway. Not just a few percent more but about 10% more throttle!! This has got to reduce pumping losses right? The manual indicates with AVLS off (no lift) swirl and tumble is hugely improved. This is why I think the ECU switches to a retarded timing map. In any case I even advanced the retarded/no-lift spark timing map quite a bit without knock to improve efficiency. I have seen good results but non of it is benchmarked and scientific. I did manage to score 40mpg with 65mph avg speed without any 'tricks', just using the throttle to keep speed. This was a good improvement and I think pretty good for a car that makes nearly 200hp.

http://i.snag.gy/3EDCE.jpg
http://i.snag.gy/UJcSa.jpg

My next steps are EGR and Lean Burn tuning. I think I'm going to induce lean burn and EGR for cruising conditions so I can run more timing and open the throttle even further. However I wanted to get some opinions here. I'm thinking a lean mixture of 15.6:1 AFR should give me best BSFC? I know some Honda lean burn engines experience crazy 20:1 AFR or Lambda ~1.4 but I'm not too sure about how it will play out with EGR and my current rather retarded spark timing map from low lift. Ie. How much timing should I add when going from stoich to such lean mixtures?

thank you for your attention.

Wow!!! Looks like your making some major progress.:thumbup:

As far as ignition timing advance goes, I think you will be the only one that will be able to tell what works for your particular car.

On my lean burn setup, on my last test run I didn't get enough seat time to really see what does what. My personal best so far at 65 mph is 38 mpg. That's with hot EGR enable and lean burn at around 30:1. I was more just testing making sure everything was working and not fine tuning for FE so I think I should be able to get over 40mpg @ 65 mph pretty easy during the winter months.

What I do is I use a long stretch of road that is extremely flat. Make runs both ways. I start with a low timing and just keep adding timing until there is no more improvement on my instant mpg readout. Also keep a close eye on the knock sensor. I then save that log. Then take more fuel away and repeat.
I also attach a MS word weather condition file with that log.
tracks « Air Density Online

Tonight's local conditions.
Spokane Raceway Park in Spokane, Washington
temperature = 33.0 deg f
relative humidity = 92 %
uncorrected barometer* = 27.98 Hg
corrected barometer* = 30.45 Hg
density altitude = 679.83 ft
dew point = 30.9 deg f
saturation pressure = 0.188 Hg
vapor pressure = 0.173 Hg
grains = 27.0
air density (w/o water vapor) = 99
air density (w water vapor) = 98

My test road around 5 miles long, very few cross roads and very flat.
http://i147.photobucket.com/albums/r...ps8facabae.jpg

Thanks for the reply. I read The 1986 Bosch handbook which says best BSFC is experienced at 1.1 lambda with PFI engines. The 2011 edition says "1.2 - 1.5" (PFI) and "up to 4.0" (DI)!

I'm not sure where I can find a road like that. Its quite hilly here. I have added about 10* of timing and set my lambda 1 to 16.7 AFR to start with.

I run this lean between 2000-3000 RPM under cruising conditions (0.3-0.6 g/rev VE) so long as in low valve lift mode. Here are the change

http://i.snag.gy/TxPYM.jpg

I will be testing this tomorrow. Oh by the way these are the non egr timings. I will do EGR after leanburn. I think the EGR will also protect the catalyst by cooling the intake charge and hence exhaust. Not sure how much hotter this AFR runs.

You can melt your heads running to lean, btw. From my limited understanding, the heat difference isn't linear and goes up exponentially the more lean you go.

Not sure leaner is hotter. This article claims CHT peak near stoich and cool off lean or richer side of it.


Mixture and CHT
http://www.whizmoandgizmo.com/Misc/EGTGraph.jpg

http://www.extraefi.co.uk/Drawings/EGT_Temp_AFR.png

Ask any private pilot with a reciprocating engine and they'll tell you which runs hotter, lean, rich...or stoich!

I wondered how long it would take to get it right. I run an EGT gauge on my metro and I'm telling you that lean mixtures as a rule burn cooler, especially at a light to medium cruise. A little on the lean side of stoich under high load runs hotter EGT.

Yeah. In any case I found throttle response and torque suffering enormously leaner than 14.5-14.6. Nearing 15s torque drops like a rock. Add as much timing you want, it won't help. On the dyno I lost about 20% torque at the wheels leaning mixture to 16.8:1 AFR with advanced timing. I found best cruising between efficiency and torque (balance) to be around 15:1. You can go leaner a bit but I really like torque and
Responsive throttle. I can just lightly press the pedal and pulse to desired speed without any downshifting or torque converter unlock. Whereas in with leanburn, the engine seems to struggle when going from lean to acceleration enrichment and causes downshifts. so I ended up with 14.65:1 dialed in for cruise. I really get good torque at 1800 rpm now advancing timing and adding a bit of fuel (14.2-14.3 afr vs 14.6 stock). i can hold speed at 1600-1800 rpm on a fairly steep hill now, not that I'm saying richer or holding speeds are better for efficiency. I still get great mileage at 50mph for a full time AWD car with 95% of the possible torque. I'm baffled how Hondas run in 1.4 lambda region without the throttle feeling like ****?

My friends brand new corolla gets 60mpg on the highway without lean burn and it is much heavier than the older lean burn Hondas as well. I think I will concentrate on other areas of engine management and tuning while meeting EPAs regulations (as a side effect).

Are you running in OL or CL when in lean burn?

Couple things I found out with my turbo lean burn Honda and now my lean burn turbo Talon.

When running in CL both cars had very poor results when in lean burn. The cycling of the O2 sensor plays havoc on the tune.

Also on both cars I did some back to back NA testing and turbo testing, and in both cases the turbo help with lean burn. In fact when running with the turbo both cars did not see a difference in torque up to 22:1.

I don't recall which thread had the info, but the only reason running lean raises temps is because it burns slower adding time in which heat will transfer to the heads/pistons/valves, etc, but lean combustion has a colder peak temperature than stoich, 14.7:1. "rich" combustion also has a colder than stoich peak temperature, the extra fuel also helps cool the engine. Part of the reason you get soot from running rich is because the combustion is too cold to burn it off.