The codes for parameters "Economy fuel" (0#01) and Consumption (0#06) both include "#", which I understand means they are calculated values. What are they calculated from? My guess is that it's from the totalized fuel consumption at each point in time. If that's correct, is it possible to obtain the raw fuel consumption (e.g., total volume of fuel consumed to that point in time)? I've written a program to calculate fuel consumed during each time interval using the 0#06 parameter, but would prefer to get the raw consumption data.
raw fuel consumption
- 3.2K Views
- Last Post 25 August 2015
jacksonmacd
posted this
16 August 2015
Administrator
posted this
17 August 2015
Yes, you suggestion is right. If most cases the base raw value is MAF Air Flow (0110), if available. That is "air consumption" the mass of air in grams per second, the mass of fuel is calculated based on stoichiometry proportion between air and fuel for the optimal burning - that is approximately 14.7
jacksonmacd
posted this
18 August 2015
OK - I will check my vehicles to see if MAF is available.
Are you saying this is the method that OBD Car Doctor uses - i.e. to convert the MAF reading into fuel consumption by multiplying by a 14.7 factor? If so, is the same factor used for all operating conditions, or does it modify the factor depending on operating conditions?
I've had to apply a "fudge factor" to my calculations based on OBD data, trying to reconcile with the amount of fuel purchased at the service station. I haven't been able to accurately reconcile the two values. That's why I had hoped there might be a parameter of actual total fuel consumed. I have seen such a parameter on the computer systems for reporting data from industrial engines (e.g., bulldozers, excavators, etc), but it sounds like such is not the case for passenger automobiles.
I appreciate receiving the new information. Thanks.
Administrator
posted this
18 August 2015
Not multiplying... divide - the mass proportion is 1 unit of fuel to 14.7 units of air. The precision of this calculation could be estimated as +- 5% or so. More precise calculation will need much more extra information as: variation of fuel-air proportion, temperature, exact density of the fuel... etc.
jacksonmacd
posted this
21 August 2015
OK - I see what you mean. I captured the MAF, then plotted it against Fuel Consumption and found a linear relationship between the two factors. Also checked the difference between the calculated total fuel consumed and the amount pumped at the service station over 14 fill-ups in my RV. Some amounts were within +/- 5%, but most are about 15%, and 3 fill-ups were more than 25% different. The average discrepancy was 15%.
From my understanding of how the vehicle's computer manages fuel consumption and varies the fuel mixture according to current conditions, it only stands to reason that the computer knows the total volume of fuel that the engine has consumed. It's disappointing that the OBD doesn't report the amount of fuel consumed.
I will put some thought into applying some other factors to modify the relationship and try to get the calculated fuel consumption closer to the amount pumped at the service station.
Administrator
posted this
22 August 2015
You are right, engine ECU "knows" more parameter that it "maps" to OBD-II, but one should also take into account that attempt to collect multiple parameters over "data bus" will takes time and thus increase sampling intervals, and as a result could reduce accuracy... From the other hand OBD-II describes direct consumption parameter (015E), but unfortunately it's very rare implemented in real vehicles.
As for the 15% accuracy, sounds to big... please check if it really "random" that is +-, but not "systematic" error. The systematic error could be due to incorrect set of fuel density or some other permanent coefficients.
jacksonmacd
posted this
22 August 2015
Yes, too many parameters doesn't work well. I loaded OBD Doctor with 8-10 parameters and found that it skipped about 50% of the readings. I've had success recording 5 parameters, but may reduce it to 4 as recommended in the instructions.
As for accuracy of fuel calculation, I always doubt my own work as a first step! Here's the logic I use in the program:
- basic reading is the OBD "fuel consumption" parameter (i.e. fuel consumption rate in litres/hour)
- omit OBD records with null fuel consumption rate
- calculate number of seconds between successive valid readings
- calculate litres consumed during time period = [consumption rate]*[seconds]/3600
- sum the litres for all time intervals
Does that sound reasonable?
In addition, there are times when the engine is running but the ODB is not collecting data (e.g. short trips, incidental time when setting up RV, etc). I've accounted for these times in the calculation of total fuel consumed, but it's an educated guess. However, my gut feeling is that it does not account for the 15% discrepancy.
The discrepancy varies from about -25% to +3% for the different fill-ups. Doesn't look systematic to me. I might be able to correlate the discrepancy to road conditions (e.g. slope, speed), but I haven't done that yet.
Dunno whether you are interested, but I am using this data in a fuel-monitoring program tailored for RVs. It does some interesting stuff about linking GPS with OBD data. I am reaching a point that I will seek beta testers. Drop me a note at rvfuelmon@gmail.com if you want further information.
Administrator
posted this
25 August 2015
Yes, all sounds correct. Looks like the observations have some systematic shift around 10%, but random is +-15 remains. I could expect possible influence sampling interval (the less, the better) and altitude (if valuable changes during the trip)... Grate thanks for your overall testing!!! Would be interesting to take and "play" with your data, thanks for the proposition, will contact soon.
