I'll test it out this weekend when I go on a 300+ mile road trip and I'll report what I find. I know the factory installed monitor on my car is usually about a mpg or two off based on my calculations. One thing that improved for a more realistic number for me when I've driven other times was changing the torque settings to "Use US Gallons" instead of imperial gallons.

-Road_wx

I do not see the USE US Gallons anywhere in Settings.

Joe

if you go to the settings under the main menu then a little than half way down there is a option to select that say "Use US Gallons"

Here is the info I got when I drove down on my trip with a 2010 Mercury Milan I4.

Built in car computer: 32.8 mpg
Torque: 31.4 mpg
Actual based off fill-up: 31.9 mpg

Distance driven 329 miles used 10.29 US gal.

Based off this info for a petrol(gas) car I'd say it's pretty accurate.

-Road_Wx

I posted originally on the old Torque forums, but I see there are some new ones now..

Anyway - I find the MPG calculations are quite inaccurate for me. (Well, the LPK rather - we use the metric system here!)

I did two long runs recently.
The first run, Torque indicated average litres/kilometre of 6.1, but actual use was 8.2.
The second run, over 660 KM, Torque indicated avg of 7.2, but actual usage was 9.

Car is: Subaru Forester 2.5L (MY08)
I do have the engine capacity set correctly in the options.

I'm having problems with the MPG (well, litres per 100k) reading myself, and would like to understand how it's calculated, since I'm guessing that it's not getting actual consumption data from the ECU?

I've got a 2007 SAAB 9-5 2.3t BioPower, and have put in all of the relevant settings that I've found in the app. I've got a cheap ELM 327 BT adapter from ebay, and it seems to be reading at least speed and RPM data correctly - haven't checked much else in detail, since I don't spend much time in the passenger seat.

When cruising at a steady speed, and re-setting the fuel avg in the car, the car was showing 8 litres, and the app was showing 6.

FWIW, I mostly run the car on E85 (85% Ethanol), which gives more power (apparently due to increased turbo pressure, ignition timing changes, etc), but also a *higher* consumption. Might this be confusing the app? I might try running it on petrol the next time I fill up, to see if that's any different.

I don't know for sure how Ian calculates the fuel consumption, but I know that it is not directly from the ECU. Simple as is the manufactures do not have to record fuel consumption as on of the parameters for OBD. Some might record fuel consumption as their own specific parameter for fuel consumption meters on the car itself, but these PIDs are usually not compatible with OBD2. The values from manufacture fuel consumption meters are usually a good bit out also!!

I have done a good bit of work in the area of getting fuel consumption from just the standard PIDs from OBD, and I could not find any simple way that would work accurately for every vehicle. The most common way is by measuring the mass air flow, and assume fuel is burnt at 14.7:1 (14.7 grams of air for every 1 gram fuel). This will only work for petrol, however the air fuel ratio is not always at 14.7, the fuel trim varies this ratio.

There has been several threads pointing out how inaccurate the average mpg figures are, getting an accurate value that works for every vehicle and fuel type is not easy at all, Torque is still a fantastic app but this is one area that it will be difficult to get right, people seem to be expecting mpg figures accurate to 1mpg, this will be very difficult to achieve in my option!

That's fair - would be good with some official information on the subject, since a lot of people seem to be asking about this, and most of us probably have no idea where the data comes from, or which data can be expected to be correct, and what's just a more or less educated guess.

When I first saw that fuel consumption data was available, I assumed that it would be directly from the ECU, and that it would be reasonably correct. I'll need to play with it a bit more, but it could be a very useful feature if you want to learn how your car behaves, and the ideal RPM for efficiency depending on load conditions, etc. Even if it's 10 or 20 percent off, if it can be expected to be consistent in how much it's off by, it'll still give a good indication of "when I'm going up this hill, for a given speed, am I best off leaving it in the same gear, or shifting down".

Personally, I would only be using the consumption figures for gauging my own driving - when I'm not driving for fun (family in the car, speed cameras, stuck behind traffic...) I might as well be saving a few quid on fuel 🙂

Measuring the fuel consumption by filling up the tank, and recording the miles you do on a full tank should be accurate to 1 or 2 mpg. One thing that could be done to improve the mpg readings in torque would be to apply a correction factor! Ie go out and measure the fuel consumption for a full tank, if you get 40mpg from the hand calculation you did yourself, but torque says it was 36mpg, then your correction factor would be 40/36 = 1.111

This is a little bit crude but should give more accurate answers for instantaneous and average mpg readings from torque.

I believe Scangauge uses this approach to adjust mpg because it asks for tank size and fill up amount after each fill up.

Quote from Timmay on February 28, 2011, 17:52
Measuring the fuel consumption by filling up the tank, and recording the miles you do on a full tank should be accurate to 1 or 2 mpg. One thing that could be done to improve the mpg readings in torque would be to apply a correction factor! Ie go out and measure the fuel consumption for a full tank, if you get 40mpg from the hand calculation you did yourself, but torque says it was 36mpg, then your correction factor would be 40/36 = 1.111

This is a little bit crude but should give more accurate answers for instantaneous and average mpg readings from torque.

So if am am following your idea you are suggesting Ian incorporate a formula to take our hand calculations and have it learn a correction factor to adjust our true MPG? If that is possible that would be great. So it learns over time to create responsible MPG information. Please Ian find a solution for this. If not it seems to me that Torque should not promote itself as being capable of tracking MPG.

My MPGs are accurate compared against the car's built in tracking (Honda CRZ) which is within 1MPG (at 45-55 MPG) of what I record from fill-ups. All I notice that is wrong, is that my MPG drops slowly in autostop, with the engine not running and MPG reads about 250 max (even though the car claims to use zero fuel with throttle closed above 1200rpm)

Maybe the vehicles which are inaccurate are not running so well, or are running open loop much more of the time?

I checked the O2 sensor volts and fuel open/closed loop status, and my car was running in closed loop all the time and the O2 sensor was at stoich - except when accelerating moderately to hard, at which point it always ran rich.

I think that some more modern engines run with a different AFR rather than 14.7 (lean burn engines, etc).

Rather than put a generic modifier into the app, I'll see if I can either calculate it, or put the AFR as a setting directly into the app settings so you can change it (slightly!) to bring the reading in like with what you would expect.

I'm not on the forums much at the moment (most of my time is taken up replying to emails since the app was featured on the market, which is cutting into development time)

Ian I think I mentioned to you before that for Diesels, you can't really related the air flow directly to the fuel flow (diesels are not throttled, the AFR increases massively at low load/rpm), but by relating it to a 2nd variable such as engine load, throttle position or RPM will give a more accurate answer. It is an area that I have looked into but didn't get as far as I had hoped due to other work I had to do.

I's think injector timing would be the best way to gauge consumption, at least for gasonline.

The pulse width is part wasted, and there is also a dependency on battery voltage. But the first can be estimated and the second ignored. Maybe the data can be checked against the measured air flow (need that for calibration anyway as injectors vary in flow).

Injector timing isn't available for most vehicles unless you have the extended PID info, though I've done some research and the load can be used to give a 'good enough' approximation of the ratio.

I have access to a diesel now and have some test code so you will see support added hopefully once I have the next version out of the way with (1.4.10 ish)

Ian

Some way of adjusting the figure would be very useful indeed.

In my case, the car itself has a display for average consumption, but it doesn't show current consumption, which is what I was hoping to get from Torque.

I agree with previous posts, that if consumption figures aren't reliable, then they shouldn't really be advertised. It's still a great app, but a bit of a disappointment if consumption figures were even part of the reason for buying an OBD2 reader + the app.

BTW, it'd be really good if it was possible to subscribe to the forums to get email replies... I never remember to check in to forums unless I'm prompted by an email.

The consumption figures should be reliable (unless you have a mitsubishi at the moment or diesel) if the displacement is correctly entered. Average MPG can be worked out several ways, and it's almost impossible to "emulate" the way the manufacturer has decided to calculate it.

The forums may well move to PHPbb or something that has less security vulnerabilities, probably running in a VM just to be extra careful. I have a new server to move all of this to, but precious little time to develop, reply to emails, read forums, and setup the new box 🙂

Ian