Joe Perez

iTrader: (8)

Regarding the tuning of a gasoline engine with the use of a wideband O2 sensor, in the presence of methanol. I'd like to expose my thinking to third party review.

The question has arisen in my mind as to the effect that injecting methanol, or a mixture of methanol and water, has on AFR.

Now, a wideband O2 sensor really measures lambda. Through the use of a scaling factor or a lookup table, the controller then converts this to AFR format, and that's what most folks are accustomed to looking at on their gauge.

With gasoline, we know that 1λ = 14.7:1 AFR. So if the engine is running at precisely stioch (1λ actual), the display gauge reads 14.7.

Now, for methanol, the ratio for stoichiometric combustion is somewhere in the neighborhood of 6.5:1. So if we were running an engine on pure methanol, we'd have calibrated our wideband controller such that 1λ produces a display of 6.5. Hypothetically, if we were to take that same engine and run it on gasoline, then a stoichiometric mixture of gasoline and air (again, 1λ), despite being 14.7:1 in reality, would still produce a display of 6.5.


Agreed so far?


Getting back to practical matters, let's say that we're tuning a turbocharged engine. Ignoring detonation, conventional wisdom tells us that we want to shoot for an AFR of about 12.5:1 for best torque, assuming optimum ignition timing. By my math, 12.5:1 for gasoline is 0.85λ (12.5 / 14.7 = 0.85). So we do that, and now we have a fuel table that achieves 12.5:1 on gasoline.

Now, let's say we're going to introduce some methanol into the engine. First off, I'm going to assume that the presence of water has no effect upon AFR, so the exact concentration of water to methanol is unimportant. And second, I'm going to assume that the quantity of water/methanol being introduced is significant enough to affect the engine's AFR in a measurable way- we'll say a ratio of 25% meth/water mixture to fuel (by volume).

So we start injecting the mixture and, assuming we do not reduce our fuel trim, the mixture starts going richer than 12.5:1. So we obviously start decreasing fuel to bring the mixture back towards our target. And this is where the big question arises.

The only data I've been able to find suggests that peak torque on meth is achieved at about 5.5:1. By my calculations, this comes out to 0.85λ, which is exactly the same number we came up with for the peak-torque lambda for gasoline.

So the question becomes: Assuming our WBO2 system is calibrated for gasoline, is it safe to assume that when the display on it reads 12.5 (equating to 0.85λ) that regardless of the ratio of gasoline to methanol going into the engine, the overall ratio of the combined mixture of fuels to air is ideal?

magnamx-5

iTrader: (7)

joe that is the million dollar question. i would assume the amalgum would be at ideal however maybe a pyrometer and some 3 way catalyst collectors would enable us to analyze it and deduce our fuel waste %. I know just from experience and general tom foolery my caloric effeciency on such and amalgum is vastly superior to that of a normal intercooled gas fueled miata. So i tend to agree with you and state my bsfc's as proof.


My first setup 5 days after i first turned the key stock ecu fuel system etc, with a 12-1 obx fpr, megan racing manis, and small 16g at 8 psi with a helper spring opening the WG so i didnt overboost while i made sure it wouldnt blow up. Stock cat and aftermarket muffler. AFR's taken from the tailpipe outlet.

mrtonyg

iTrader: (1)

Joe, most WBO2s actually measure in a voltage range of 0-5v. Stoich (Lambda) and AFR are the conversions. Stoich and lambda are basically synonyms.

Let's say an engine is running with the WBO2 putting out 3.5v. In the look up tables for the controller, 3.5v equals 16:1 AFR. Then that engine that produces 3.5v from the WBO2 has an AFR of 16:1 regardless if it's burning alcohol, gasoline, propane or any other fuel. The stoich ratio (lambda) will change depending on the fuel used because each fuel has it's own "sweet" spot.

Did I misread what you were trying to state?

j_man

Yup, you're absolutely correct. The WBO2 measures lambda. And you tune for that. Some say the Methanol makes max power at a bit richer than 0.85 lambda though. That should mostly matter if you're running the engine on just methanol. For intake injection, just tune for 0.85 lambda on the gauge regardless whether you're spraying methanol or not

P.S. Btw the water is supposedly changing the O2 sensor readings, but so minor that you can ignore it

j_man

He was trying to state that one should ignore the type of fuel he is running - just gasoline, or gasoline + methanol injection, but just tune for the same reading on the O2 sensor (== lambda) and ignore AFR and all that crap. And he is right.

pdexta

iTrader: (17)

I think that what Joe is saying is that the wideband does not actually read an AFR, instead it reads the exhaust gasses distance from lambda. The wideband then converts that number to an air/fuel ratio that applies to gasoline. When you put a different fuel in the motor the AFR (that was calculated for gasoline) is now incorrect.

If the gauge was reading a true AFR guys running a 4:1 meth injection should be targeting something around a 11.1 air fuel ratio for peak torque (4 parts gas @ 12.5, 1 part meth @ 5.5). Since the gauge is actually reading the exhaust's distance from lambda peak torque will be seen at 12.5:1 regardless of the fuel that the motor is running.

(Hopefully I got that correct, please correct me if I'm wrong)

sixshooter

iTrader: (12)

The zirconium dioxide oxygen sensor changes the flow of electricity through a circuit based on the concentration of oxygen in the exhaust gasses. Since all it determines is oxygen concentration one must determine if the presence of methanol increases or reduces the concentration of oxygen in the exhaust gasses. The stoich value of burning methanol is irrelevant to the sensor, because the exhaust created by stoich burning methanol may have a higher or lower concentration of oxygen than that of the exhaust of stoich burning gasoline (I'd look it up right now, but I'm driving on the interstate). This difference (and there likely will be one) must be factored into the calculations regarding the computation of fuel ratios based solely on exhaust gas oxygen concentrations.

y8s

iTrader: (8)

which is why you should just learn lambda and know that it is a direct ratio of air to fuel based on complete combustion as "lambda=1".

so if you're cruising with some random fuel, you know you're good when your gauge says "1ish" or when you're under power, you know you want to be <1. and when you lift, you want >1.

Joe Perez

iTrader: (8)

I believe that you are mis-interpreting how a wideband system works.

And I'll explain why I believe that.


In the configuration software for the LC-1, there is a screen where you tell the system what fuel you are using. A screenshot is attached:



Note that it says "... to calculate AFR from Lambda", suggesting that the sensor itself responds to lambda.

In the documentation for that section, of the software, it states the following:

On this page you can see the software version of the LC-1 and you can change the multiplier to
calculate AFR from Lambda. A number of different multipliers are already pre-selectable but you
can change it to a custom one for the particular fuel you are using.

From this, I infer that the LC-1's sensor element responds to lambda (which would be a constant for the stoichiometric ratio of any fuel) and then relies upon this setting to determine the correct multiplier it will use to convert that lambda value to an AFR value for display purposes.



The AEM gauge is much less configurable, and the documentation is not quite as clear. However it does state that "UEGO sensors use a “current pump” to determine the actual oxygen concentration within the sensing element" and in the "specifications" section in the back, it indicates "Measuring Range: 0.751 to 1.143 Lambda". Additionally, it notes that of the five switch-selectable calibrations, four are intended for use with gasoline, and implies that if an alternate fuel is used, the fifth calibration, lambda, should be selected. It then lists a table of the 1λ values for various fuels, along with a table showing what AFRs corresponds to the range of 0-5v output for all those fuels. In every case, the 1λ point for each fuel occurs at 2.35v, despite the fact that there is no user-selectable calibration for discriminating between, say, ethanol, methanol, propane or CNG. One is simply presumed to have selected the lambda display calibration for any non-gasoline fuel.


So I believe it to be a safe assumption that, like their nernst-cell cousins, wideband O2 systems natively respond to lambda. IOW, for the ideal stoichiometric ratio of any fuel to air, the sensor itself will detect 1λ, and the controller, assuming a linear 0-5v calibration, will output 2.35 volts. This would be true for gasoline at 14.7:1, methanol at 6.5:1, ethanol at 9:1, propane at 15.7:1, etc. Any of those combinations is 1λ, representing neither a surplus nor a deficit of free oxygen in the exhaust. It then falls upon the controller to perform an arithmetic operation to convert that into an appropriate AFR, which is done solely for the purpose of driving the user display. Were it not for the display, AFR could be ignored entirely and everything done in straight lambda.


Perhaps it would be easier to think about this in terms of a narrowband application, where the actual sensor itself directly produces an analog voltage, with no intermediate conversion steps. If you have a narrowband sensor in an engine running gasoline, and the mixture is exactly 14.7:1, then the voltage output of the sensor will be midrange, approximately 0.5v. If you then convert that engine to operate on 100% methanol, and you set it up so that you're running at exactly 6.5:1 AFR, then the output of the sensor will still be 0.5v. This is because both conditions represent perfect stoichiometric combustion, at exactly 1λ.The narrowband sensor has no concept of AFR as such, only lambda.

From all of the data that I can find, it appears that a wideband sensor (again, the physical sensor itself) operates in essentially the same way, responding to lambda, with the subsequent translation of lambda to AFR done by the microprocessor inside the controller, based upon a pre-determined conversion formula which is specific to the fuel being burned.



What I don't know is whether this validates my belief that, assuming an "ideal" (meaning best power) AFR is 12.5:1 for gasoline, and 5.5:1 for methanol, both equating to 0.85λ, does it follow that tuning for a displayed value of 12.5:1 on a wideband system calibrated for gasoline, which we presume to indicate 0.85λ regardless of the specific fuel involved, is optimal for any blend of two fuels (gasoline and methanol) at a ratio that is not precisely known.




Well, yeah. I think that's the basis of what I'm postulating. Not so much that AFR should be disregarded (as it's still how my brain wants to work when thinking of mixtures) but rather that, based upon the system's tendency to translate any mixture which achieves 1λ (regardless of the actual AFR that 1λ represents) into a displayed value of 14.7, that using that system of scaling is reliable for any mixture of fuels where the lambda value for best power is the same for each fuel in the mixture (0.85λ) regardless of what the ratio of fuel A to fuel B is.

Joe Perez

iTrader: (8)

Well, in a nutshell, I'm essentially trying to confirm that what you've said, which is also what I assume, is in fact true.

Let's say that I do have my gauge set to display lambda rather than AFR. If we assume that MBT for gasoline is 0.85λ, and MBT for methanol is also 0.85λ, is MBT for a mixture of gasoline and methanol (and water) at an unknown ratio also 0.85λ, and that while tuning a turbocharged engine running said mixture of fuels, one should be striving to achieve 0.85λ?

y8s

iTrader: (8)

that's what wikipedia says...

Joe Perez

iTrader: (8)

Wikipedia says a lot of thing.

wes65

iTrader: (7)

Joe, you've done it again. My mind is officially blown. I see what you're saying and I am confident to go leaner with my mixture. Before, i was thinking that since stoich was 6.5:1 for methanol, I was going to be shooting for somewhere in between MBT AFR for methanol and MBT AFR for gas. That is not the case.

y8s

iTrader: (8)

indeed. but it corroborates what you thought and what I thought before looking at wikipedia.

Joe Perez

iTrader: (8)

You're the one who started me down this path with your 7 GPH nozzle. You've been here long enough to know that you can't do something unconventional without somebody turning it into a dissertation.

wes65

iTrader: (7)

Haha, very true. Although, I would say around these parts, you're the king of dissertations.

mrtonyg

iTrader: (1)

You are right, I will just blame it on trying to answer a post like this at 1:00am!

ZX-Tex

iTrader: (15)

I think the only reason that the settable value is there for the LC-1 is for the purpose of the scaling factors in the analog outputs. You can set up the scale on the output using AFR or lambda; If you select AFR, then set it to read 0V at X.X AFR and 5V at Y.Y AFR, the value input in the drop down menu scales it into the equivalent lambda value. I think if you just set the outputs in terms of lambda, then it is insignificant. For example I set my output that goes to the dash gauge with lambda values. IIRC, 0.5λ = 0V, 1.5λ = 5V. I think in that case it ignores the value in the AFR multiplier.

Also the dash gauge is really just a volt meter, scaled to read for gasoline AFRs. You could just replace the scale on the gauge to read in λ instead of AFR. Or, like someone said above, know that a gauge reading of 14.7:1 AFR is λ=1 regardless of what fuel you are running. It is after all, like someone said, just measuring residual oxygen levels.

j_man

Yours is just a voltmeter I guess but the good ones are not. My TechEdge gauge communicates with the controller (serial RS-232) and displays the lambda the controller internally is reading (which excludes any errors caused by analog conversion, ground diffs, etc.). And you can configure the gauge to convert that lambda in whatever else units you want.
The good Innovate gauge is the same way - serial connection to their wideband.
I use the gauge to make sure the ECU is properly/precisely calibrated for the wideband. I.e. my AEM EMS needs a 0.96 scale (4%correction) when connected to a TechEdge WBO2A unit.

ZX-Tex

iTrader: (15)

The LC-1 connection to my Adaptronic is via RS232. I like that because it removes a source of error.