Alpha-N Fuel Consumption Math

[f0xy]

I'm looking to measure fuel consumption as the engine is running. I already have the hardware in place to calculate this, but do not know the math. With the current ECU setup, I think I have access to all the variables I need to keep a running total of fuel consumed, since the engine has started. This engine is 6 cylinder, mapped alpha-N (ITB), with fixed fuel pressure of 3.5bar (51psi). The injectors are Bosch 0280155715 which flow 296cc/min @ 3.5bar. I have the following data available directly from the ECU (and most other things if required):

RPM
Injection Time (ms)
Fuel Rail Pressure (psi) [if required for any slight variance]

I'm not looking for high accuracy, just something near would be great, as fuel level isn't of much use in this application. I'd rather know how much has been consumed, as generally it always starts with a full tank.

Does anyone think this is a viable thing to calculate with an equation?

Thanks.

[Benrad]

The main trouble you'll have is that the stated flow rate will be at full injector opening, so error at small fuelling quantities will be large (as the injector spends more of each injection event, maybe all of it, at part opening)

Secondary might be how many figures are stored for the injector opening time

I'm both cases because you're multiplying that small error many times over you'll end up with a large error at the end. Perfectly doable but don't trust it if you've been driving gently and always treat it with suspicion

You'll need:
Injector opening time * stated flow rate * constant to take into account units * engine speed (rpm) * 6 / 2 assuming it's a four stroke engine

[f0xy]

Thanks, yes totally agree its not going to be perfectly accurate, its just to get some sort of idea about what is being used.

Sample rate - I'm only currently logging to file @ 10hz, although the serial feed from the ECU is much quicker @ 115200 baud, and if need be can log to file at 100hz+, but I don't need all that in the file. The calculation will run 'on the fly', so to speak, so I think the communication/output from the ECU is quicker than i'd ever need at my maximum RPM.

This is a car used on track/racing only. so spends most of its time at wide open throttle. I know it will complete a 45 minute race on a full tank (55L) without issue, but I'd rather have an idea of how much is left without having to pump the tank dry and check every time.

I currently have a couple of log files, which I can simulate the equation/math in, albeit recorded at a lower frequency - I know approximately how much fuel was used in each session the log file(s) are from, so may be able to see how feasable it is

[anonymous-user]

At a first approximation, ingnorig injector slope and delta pressure compensation, fuel consumption can be estimated by simply integrating the injector opening duration and engine speed.

For example, once per second you get a value for engine speed and injector duration, lets say 2000rpm and 12 m/s and assume you have a 4 cylinder engine.


So we first consider those single sampled values to be representative of the "average" consumption for that second.

Total number of injection events over that one second period is:

rpm divided by 60, divided by two (four stroke) multipled by number of cylinder = (2000 / 60 / 2) * 4 = 66.6 events per second


Total injector duration is number of injection events multipied by the length of each event = 66.6 * 12 = 800 ms = 0.8 seconds

Total fuel flow is injector flow vs total time (lets say your injectors flow 300cc/min, which is 5cc per second) 0.8 * 5 = 4cc "consumed" in that particular second.

If you add up the flow average for each second in a rolling total, you will summate (integrate) the total fuel consumed.


You get more accuracy by getting faster updates. Get data once every 100ms (ie 10 times per second) and your total will be more accurately reflect real consumption.

Obviously, we have ignored fuel injector delta pressure (if you have an inlet manifold referenced pressure regulator then injector dP is held constant) and effects caused on short duration injector events (< about 2ms, where there is not time for the injector to fully open, hence it flows less)

[Benrad]

If it's for on track then accuracy should be reasonable. Plus you've got the tools you need to check your accuracy (do pump outs on the first few outings).

When helping run a modern race car last year I was pleasantly surprised by the accuracy the ECU gave for fuel used, but they had more information than you do to make it accurate of course. Plus we pumped out every time anyway. The mechanics would test my mental arithmetic by getting me to tell them what they were going to pump out at the end of test days where we'd been randomly throwing fuel in the thing, but I'd be close when after a whole day!

Nothing to lose by giving it a go, so let us know how you get on

Max_Torque has put some working behind the answer I gave but we agree on what you should be doing.

[f0xy]

Thank you both for the explination and information. So I've adapted it for my setup, logging at 100ms intervals, does this make sense? (Using your numbers)

(2000 rpm / 600 / 2) * 6 cylinders = 10 events
10 events * 12ms = 120ms injection time
120ms * 0.005cc = 0.6cc consumed in that 100ms

Thats if i've understood your post correctly. I will apply that to one of my log files shortly if so!

Cheers

[GreenV8S]

Are you running constant fuel pressure relative to manifold pressure?

[f0xy]

No, with the ITB I run no reference. This is fixed fuel pressure using a 3.5 bar regulator. I have a fuel pressure sensor on the rail also, and generally see 51psi +/- 1psi all the time in the logs

[GreenV8S]

The injector flow rate while open will vary with manifold depression. I think you need to take that into account in your calculations, but it wouldn't be easy given that you are not using MAP. Would you consider either adding a (damped) MAP sensor to one of the runners, or adding a flow rate sensor in the fuel supply and return lines?

[f0xy]

GreenV8S said:

The injector flow rate while open will vary with manifold depression. I think you need to take that into account in your calculations, but it wouldn't be easy given that you are not using MAP. Would you consider either adding a (damped) MAP sensor to one of the runners, or adding a flow rate sensor in the fuel supply and return lines?

Previously I ran MAP on this engine (all 6 runners into a small vacuum chamber). It worked OK but wasn't the most stable reading as you would expect. This was only in place when the car had servo assisted brakes, as it ran off the same chamber). It was removed when the engine was moved to a different chassis with pedalbox, etc.

In an ideal world it would be great to have flow sensors on either end of the rail, but I can't imagine fuel flow sensors are cheap?
Considering this isn't a critical measure and I'm only using it for reference to get some sort of idea about whats happening, I don't want to invest heavily in knowing the exact consumption really.

I put the calculation above on to one of the recent log files. This is a single lap of Cadwell Park, 1,40, and would suggest around 1.2L of fuel was consumed. Not sure how realistic this is (yet) though.

[anonymous-user]

GreenV8S said:

The injector flow rate while open will vary with manifold depression. I think you need to take that into account in your calculations, but it wouldn't be easy given that you are not using MAP. Would you consider either adding a (damped) MAP sensor to one of the runners, or adding a flow rate sensor in the fuel supply and return lines?

For a race car (i assume this is a race car) then the throttle should only ever be wide open or fully shut, and of course, when fully shut, fuel consumption is low, so the wide open portion will dominate. And the flow through an orrifice is proportional to the square root of the pressure difference, and with 3.5bar static fuel rail pressure, intake pressure changes by lets say 0.7 bar, which is 20%, so if we include the deltaP, that just a 14% change in flow rate.

Direct flow measurement is difficult with a return system, and typical gear type flow meters cost around £300.

[anonymous-user]

f0xy said:

I put the calculation above on to one of the recent log files. This is a single lap of Cadwell Park, 1,40, and would suggest around 1.2L of fuel was consumed. Not sure how realistic this is (yet) though.

Sounds about right, 1.2L of fuel, over say a 1 min 40 sec lap, is an average power of 408kW, so assuming your engine is 25% efficienct, your average engine power over a lap is 102 kW (137 bhp)



Assuming you refuel the car from jerry cans, take some household weighing scales with you next time, and note the mass of fuel you put in and when you refuel (weigh jerries before and after each refuelling, difference is fuel added to the tank, assuming you haven't spilt it all over the floor!), then you can compare those values to your post processed calculated numbers. You could probably come up with a "correction" factor for any particular track to get your calculated figure closer to the actual consumption

[f0xy]

Max_Torque said:

Sounds about right, 1.2L of fuel, over say a 1 min 40 sec lap, is an average power of 408kW, so assuming your engine is 25% efficienct, your average engine power over a lap is 102 kW (137 bhp)



Assuming you refuel the car from jerry cans, take some household weighing scales with you next time, and note the mass of fuel you put in and when you refuel (weigh jerries before and after each refuelling, difference is fuel added to the tank, assuming you haven't spilt it all over the floor!), then you can compare those values to your post processed calculated numbers. You could probably come up with a "correction" factor for any particular track to get your calculated figure closer to the actual consumption

Thanks, yes, I will stick with that calculation for now and monitor when its next out in a couple of weeks. I do use cans, so will attempt to do as you suggest. For reference the car is 280hp and weighed near enough 1190kg in the log above with the fuel load it had.

[anonymous-user]

f0xy said:

Max_Torque said:

Sounds about right, 1.2L of fuel, over say a 1 min 40 sec lap, is an average power of 408kW, so assuming your engine is 25% efficienct, your average engine power over a lap is 102 kW (137 bhp)



Assuming you refuel the car from jerry cans, take some household weighing scales with you next time, and note the mass of fuel you put in and when you refuel (weigh jerries before and after each refuelling, difference is fuel added to the tank, assuming you haven't spilt it all over the floor!), then you can compare those values to your post processed calculated numbers. You could probably come up with a "correction" factor for any particular track to get your calculated figure closer to the actual consumption

Thanks, yes, I will stick with that calculation for now and monitor when its next out in a couple of weeks. I do use cans, so will attempt to do as you suggest. For reference the car is 280hp and weighed near enough 1190kg in the log above with the fuel load it had.

280 bhp is the PEAK power of your engine. The average power over a lap, which includes braking and cornering is a lot lower!