Calculating CFM from Throttle Position and Manifold Pressure

[Sonreir]

Hey all,

I'm a bit new to the game in the process of doing some research on fuel injection. I while away my working hours on software programming, but engines have always fascinated me and I like to tinker in my spare time.

The problem I'm currently facing is that I'm unsure on how to calculate how much air an engine is using at any given time.

I'm working with a fuel injection system from a Kawasaki Ninja 650 (EX650) and have a throttle position sensor and a MAP sensor. The throttle bodies are 38mm in diameter.

I know that using these two sensors I should be able to arrive at a CFM number for air usage, but I'm a bit stuck on how to continue.

What I have so far is the cross sectional area of the throttle body as it relates to throttle position in radians (attached image).

How do I use this to arrive at CFM?

Thanks in advance,
Matt

 

[ivymike]

Knowing pressure drop (across the throttle) and throttle restriction, you should be able to calculate volume flow rate thru the restriction to give that pressure drop.

http://www.engineeringtoolbox.com/duct-friction-pressure-loss-d_444.html

one problem is that you don't know how restriction varies with flow and throttle position for your valve. You could try assuming that it's linear, but it's probably not quite.

Another question to answer is "where do you want to know CFM" as volume flow rate will be a different number in different places in the system, depending on the density of air at those places.

 

[Sonreir]

Thanks for the link! Time for some algebra.

Ideally, I'd like to know (or reasonably assume) the CFM actually ingested by the engine. How is it that CFM changes? My understanding is that in a closed system, only the air velocity would change and flow would remain the same (after taking into account all restrictions)?

 

[dgallup]

You need to convert to standard CFM or to mass so you can calculate the appropriate fuel mass to inject.

It is my understanding that most systems use the Map sensor at small openings and then switch over to the TPS at larger openings (I may have that backwards). Evidently getting a good cross over is quite tricky. Generally speaking, engine control software is pretty complex. You may want to look at the MegaSquirt site, I believe the software is open to see.

----------------------------------------

The Help for this program was created in Windows Help format, which depends on a feature that isn't included in this version of Windows.

 

[Lou Scannon]

You need an accurate model of the throttle effective area vs. position. Effective area is not the same as projected area.
The volumetric flow of a given engine is a complex function of the respective manifold pressures, the speed, and various temperatures [which are used to model the increase in charge temperature as it enters the cylinder].
Looking at your task, at some point you will need to convert from CFM to SCFM. Since SCFM relates to standard pressure and temperature, it is really equivalent to mass flow (or some equivalent, such as trapped air mass), which is what you need to know in order to control fuel injection. The conversion to SCFM is actually dead simple, so don't let that worry you. But you will need to know temperature as well as pressure upstream of the throttle, along with the throttle deltaP.
This is all pretty basic stuff and based on your questions, I think you really need to start with a textbook. I haven't consulted it lately, but the Bosch Handbook is pretty simple and to the point, and might be an inexpensive option.


"Schiefgehen will, was schiefgehen kann" - das Murphygesetz

 

[BrianPetersen]

I have another question: why are you trying to do it that way?

I have some familiarity with Kawasaki fuel injection systems, and they have two separate but connected sets of maps in the ECU, one using engine speed and throttle position (alpha), the other using engine speed and intake manifold pressure (MAP) relative to barometric pressure. The ECU makes no attempt to "calculate", it simply establishes the current RPM (actually, it uses the inverse, the number of CPU timer pulses for the crank sensor to make one revolution, but I digress), and at a certain consistent crank position, it samples the throttle position and the MAP sensor. Then it looks up the target fuel delivery based on N-alpha for one set of maps, and based on N-MAP for the other set of maps. It is a table look-up operation, not a calculation - aside from applying corrections for barometric pressure and temperature to the numbers that it obtained from looking up the base table.

At light engine load (below a certain RPM and throttle position combination) it uses predominantly the N-MAP map and at high engine speed it uses predominantly the N-alpha map. This is because very tiny throttle position changes at low RPM result in large changes in MAP so it is better and more accurate to use the MAP under those conditions rather than trying to interpolate alpha. But, both the N-alpha and N-MAP maps are complete over the entire usable RPM and throttle position range. This is so that if either the throttle position sensor fails or the MAP sensor fails, the engine can continue to operate using the other sensor values (although with the MIL "check engine" lamp on).

To my knowledge, no automotive or motorcycle EFI system attempts to "calculate" airflow from engine speed and sensor values. They all do it via table look-up. It's the job of the powertrain calibration engineers to establish what the values in that table need to be in the production vehicle. Obviously this is done by dyno testing, and obviously there needs to be a "guess" map to start off with, to allow the engine to start and run so that further calibration can be done ... but that "guess" map need not be all that accurate, so there is no sense to over-think it.

So what is it that you are trying to do, that goes outside the usual model described above?

 

[Sonreir]

Well... sort of...

I'm not planning on running the throttle bodies in conjunction with the EX650 ECU (or even on the same bike). I'm just after the hardware.

I was hoping to generate my own maps and so I need a better understanding of the math before I can do down that road.

So my processor won't be calculating at all. You're correct in that I'll be using maps, but the maps that came with the hardware probably aren't going to work for this application.

Rather than go through trial and error with a wide band sensor and every possible permutation of throttle position and engine load, I was hoping to generate something close and then trim it as necessary (based on MAP sensor readings and changes in throttle position).

 

[MikeHalloran]

A while before digital ECUs went into production, there was a flurry of activity with electromechanical self-adaptive controls, that decided what to do next based on how the engine responded to what they did last. Basically they changed afr or ignition timing or both, and monitored the response using the flywheel teeth as a tone ring tachometer, effectively using the engine itself as a transducer.

There were technical papers and patents, some assigned to Harmon Electronics of Grain Valley, Missouri, USA. The published material may be illuminative. I suspect that many of its lessons were adapted and re-used in more recent ECUs equipped with learning algorithms.



Mike Halloran
Pembroke Pines, FL, USA

 

[LionelHutz]

If you really think you need to do this then I'd be looking up build it yourself flow bench information and using that to test the throttle body and associated intake components.

Otherwise, stick a wide band O2 sensor into the exhaust and write up a self learning algorithm. Use a MAP vs rpm vs desired AFR table and let the engine self tune to meet those desired AFR numbers.

 

[decipha]

the most accurate way is to not calculate it at all, simply measure the airmass entering the engine with a mass air flow sensor

the 2005+ ford maf would be most ideal for what your trying to achieve

simply feed it a 12v and 5v reference and measure the voltage output

The airmass will vary dependent upon the diameter of the pipe you put the meter in, just calculate your voltage to airmass transfer function and your done

the beauty with measuring airmass is that temperature, barometric pressure, air density, etc.. aren't a concern