Interesting results - shifting into 5th early vs. late

[JHZR2]

Hello,

When accelerating, I typically shift as early as possible without lugging the engine. Lately I have tried a new technique - shifting according to the turbo gauge on my 04 saab 9-3 and the MPG gauge on my 91 BMW 318i.

The cars:
2004 saab 9-3 2.0L low pressure turbo, MT, running Mobil 1 0w-40 oil, stock tires at 41psi front, 38 rear (OE inflation levels)
1991 BMW 318i, 1,8L MT, running Shell Rotella 5w-40 oil, stock tire size at 32 front, 35 rear (increased over stock to just below the point where they feel like a buckboard).

Ive always logged every tank of fuel. I dont believe that any results shorter than an entire tank of fuel are useful in any way, and even single tanks fall under the stipulation that "one test is no test".

There are lots of ways to shoot holes in this test, but here is what I did: Two tanks each (two is nearly one and thus is not statistically significant in any way, but at least it starts to show some trend). I alternated drivign styles, one tank shifting as early as possible (usually shift around the 2000 RPM point, from slightly over to slightly under), the other keeping the gauges maximized (turbo gauge on the cusp between vacuum and boost, the NA engine just as high an instantaneous reading as possible).

The saab is my highway only vehicle. Ive typically gotten between 36-38 MPG with it. Its typical driving profile is 35 miles of straight highway between cold starts. The BMW is my commuter car, and sees 8.9 miles of city and mixed driving (multiple lights) between cold starts. Both cars see two trips over a large bridge every day (well I only drive one of the two cars each day, but you know what Im saying ). I don't ever shut the car off, but I do take it out of gear and coast when possible. I also always run the A/C.

So, results are:
tank 1 (shift early):
saab = 36.27 BMW = 25.69
tank 2 (shift based upon the gauge / hold lower gear longer):
saab = 37.92 BMW = 27.71
tank 3 (same as tank 1):
saab = 35.96 BMW = 26.15
tank 4 (same as tank 2):
saab = 37.64 BMW = 28.01

So it seems that the technique of shifting merely at the point where you don't lug to get into a higher but more "efficient" gear is not necessarily correct (I had read one study from the early 1980s that suggested that mild lugging is OK, and that even skip-gear shifting should be routinely done to maximize economy!). Granted, this is only 2 tanks each (about 300 miles each for the BMW, about 500 miles each for the saab), but perhaps it is insightful. These are pretty different engines... one is SOA 1991, the other is SOA 2004. Both had arguably the most advanced engine management systems of their time. I was sure to keep the tires at their setpoints throughout the slight changes in winter. I assume based upon the data that all tanks were with a 'winterized blend', as there doesnt seem to be deviation based upon that (all fuel for this test was bought at the sunoco on rt 130 in Collingswood, NJ).

Ill re-do this test in the spring when temperatures even out. Doing more of the same now wouldnt be worthwhile because I installed snow tires on both cars, I changed the oil in the BMW to schaeffer's 5w-40, which has a higher HT/HS rating, and so will be a bit thicker at the bearings, and also, because I noted a crack in one of the breather hoses under my intake on my BMW - which never set off a CEL, or caused an unstable or not-smooth idle, but may have leaked air, causing higher consumption. Ill try to do the whole test, alternating driving styles, over the whole spring/summer/fall, when snow tires and whatnot arent in the equation.

So, any thoughts??? Its only two tanks with each style, so it is effectively not a real test. But a trend is seemingly forming. It certainly could be all in the noise - this winter has been odd, with pretty wild temperature swings, which can have an effect. I was always under the impression that shifting up as early as possible is preferable, but it seems that at some extent, for both a turbo and a NA engine, that this isnt the case.

Just what I found... YMMV

JMH

[phoebeisis]

Do you have a scangauge or any way to get an instantaneous mpg reading.If you do,you can check you MPG at near lug speed in 5th,and check your mpg in 4th at the same speed.Which one is better?
I have noticed that my scangauge shows better mpg at 55 mph than 50 mph.The trans-AT- locks up at ~40 or so.I always assumed that the lower speed would get me better mpg, but apparently it isn't so.The motor must have an efficiency "sweet spot" slightly higher up the RPMs that more than counterbalences the extra fuel used at the higher speed.
Maybe your motors have similar "sweet spots."
Like you said; it will be very tough to get a feel for the actual mpg differences from just 2 tanks-especially since the difference you are trying to measure is just a few seconds per run thru the gears .You run it in 4th just a few more seconds;hard to measure that difference over the background of all the "normal" steady 5th gear driving.
Luck,
Charlie

[JHZR2]

I dont believe any results from a scangauge. An instantaneous reading in a test that has SO many variables is hard to deduct a trend from, without putting a lot of attention into it, thus taking attention away from the driving. That said, the BMW has an instantaneous MPG gauge (the saab only has average), and Ive found that I tend to stay in the higher MPG numbers longer when holding the lower gear a bit longer. Similarly, shifting arly (say 50 MPH at 2000 RPM in the saab) causes the engine to require boost to maintain speed, vs. 55-60 mpg in 5th gear, where my vac/boost gauge stays right at the zero mark.

Your comment about sweet spots is well noted... I wouldnt doubt for this to be the case, and could potentially even have smething to do with the ECU programming. A few folks that own trucks like mine (I have a 98 Chevy S10 ZR2) have noted that cruise at 70 mph yields lower fuel economy than at 67 or 72...

Next time around Ill put the scangauge on the saab (the bmw is only OBD1), and see if I can deduce any instantaneous trends. Though Ill not trust them per se, I can at least see if there is anything obvious...

JMH

[koreberg]

I think instantaneous FE would be more valid than tank based FE. You're trying to measure the difference between 30 second driving events. But you're using the culmination of 10000s of driving. Even if you have 100 of those 30 second events, there are still so many other variables that could affect your FE that are un accounted for.

It would be much more valuable to find a deserted road and accelerate to your city crusing speed over and over using the 2 different ways, and watching instantaneous FE and setting your trip timer etc. Going both ways and driving the different ways on the same parts of the road while temperature is pretty much the same would elminate most of the external uncontrollable conditions.

[mcurran07]

I don't quite grasp your method to shift according to the gauges. You say you kept the gauge readings in a range, but when did you shift and to which gears, and what throttle percentage etc?

[jcp123]

That's an interesting experiment. This is why when I came here I thought keeping my SVT Focus' revs at around its torque bump at 2200rpm would be highly advantageous - as its torque increases, the load should decrease. Oddly enough, though, I've found that shifting at around 1800 has produced the best results for me. I may try holding 5th longer now, though, and hitting 6th as I did before at 2000-2200rpm.

[koreberg]

I almost bought 1 of those svt focus. But then determined I would rather wait until I could afford something bigger. I really wanted a g35 or rx8. I am slowly changing my mind on that now.

What sort of FE do you get from that car? How does the sc affect the fe for that car?

I also am not sure when you're shifting in your second test. Seems like you're shifting at around 2k in your control test. When are you shifting in your other test?

[JHZR2]

Originally Posted by koreberg

I think instantaneous FE would be more valid than tank based FE. You're trying to measure the difference between 30 second driving events. But you're using the culmination of 10000s of driving. Even if you have 100 of those 30 second events, there are still so many other variables that could affect your FE that are un accounted for.

It would be much more valuable to find a deserted road and accelerate to your city crusing speed over and over using the 2 different ways, and watching instantaneous FE and setting your trip timer etc. Going both ways and driving the different ways on the same parts of the road while temperature is pretty much the same would elminate most of the external uncontrollable conditions.

I respectfully disagree. Think of it this way... say Im driving down a slight decline... that I cant really notice from my point of view as the driver. All of a sudden, I have a 5 mg bump, and I could mistake it for something that I did. Ditto for an inclne, headwind, tailwind, etc. And lets not even go into latent thermal energy in the engine, slight changes in oil viscosity, temperature, etc. So, all those "unaccounted" things that you claim are so present in the tank scenario, are more present in the instantaneous scenario. The difference is that over a tank, various conitions ought to ultimately average out (maybe not, granted, but at least there is more repetition), and so behavoral changes due to technique can make a real effect to the overall average.

I agree about doing it on a deserted road, where most everything is the same, all the time, and the number of runs were done at once, so there are consistent conditions, repeatability, etc. but I dont hae that luxury. That would be the ultimate, and where the SG and a datalogger would make the data MOST relevant.

JMH

[JHZR2]

Originally Posted by mcurran07

I don't quite grasp your method to shift according to the gauges. You say you kept the gauge readings in a range, but when did you shift and to which gears, and what throttle percentage etc?

Typically I shift as early as possible. This makes my RPMs hover around 2000 RPM (so, say I shift at 2100, it drops to 1800 rpm, I accelerate to 2100, whift again, etc). I shift to avoid lugging, at as low an engine speed as possible. This is the old get into overdrive as early as possible type of technique.

The saab has a boost/vac gauge, the BMW has an instantaneous MPG gauge. In the saab, I drive such that the needle of the boost/vac gauge is always near the 'neutral' or zero point. This changes the rpm shiftpoints and speeds higher. In the BMW, I watch the instant MPG needle and let it reach a maxima, not shifting until its value starts to drop (too high of RPMs for the speed Im going at and the mechanical advantage gained by the gears). This also causes me to drive at higher RPMs in a lower gear/a.k.a. higher speeds in a lower gear than I would have normally.

Granted, its not severe... it doesnt have me doing 70 pmh in 4th gear at 4500 RPM... Im talking about cruising in 4th at 50-55 MPH, as an example... prime territory usually for me to be cruising along in 5th, since the engine doesnt lug and can accelerate out of it if I need to.

FWIW...

JMH

[kngkeith]

The scangauge will help monitor variables in your behavior too, to help control the test. TPS comes to mind.
Does the boost kick in only when you try to shift early to 5th? Load varies at different speeds. Perhaps progressive shifting- shift early in low gears, adding rpms as you add gears. Similiar to what JCP123 wants to do in the Focus. Please continue your experiment now, the sample size will help to control for other variables.
Keith