BSFC Mapping
 

How would you go about it and what gauges would you use to map it?

Been thinking about it and not sure which gauges on my UG will be best.
Relative throttle position/absolute position
Load
MAP

I think you'd need a dyno that can hold the engine at a constant rpm and vary load.

In addition to a good dyno that will let you impose a specific load on the engine, you will need two fuel flow-measuring devices--very sensitive ones. One measures the fuel coming to the engine from the tank, the other measures the fuel going back to the tank. The ones with the sort of precision you use are quite expensive, last I heard.

BSFC is fuel used per time per unit of power (or per unit of torque or per unit of pressure in the cylinder). So you have to get quantities for all three of those things--fuel used (generally by weight), time, and power or torque developed.

You can approximate BSFC for some specific points in the load/power regime if you get readings of gallons/hour of fuel use (the UG will approximate that from fuel injector duty cycle) and power or torque currently being developed. (Your UG will, if it provides those numbers at all, be approximating them from ... probably from a whole lot of inputs frankly.)

So you get approximations multiplied by approximations divided by approximations. Which means any results you get are going to be a little bit better than total wild guesses, but only a little.

-soD

So, since the new UG has HP, torque and kilowatts, could you combine those with VE, Load, and whatever other gauge and map it?

If you have a fast sampling rate, you should be able to calculate it from logging the engine sensors through the OBD2 port.

The HP is an approximation. The torque is also an approximation. Ditto the kilowatts (which is power just like HP, but in a different scale). VE is somewhere between an approximation and a wild guess.

Using those to calculate BSFC means you will have numbers with several sets of approximations all multiplied together. The results will be little if any better than picking numbers out of a hat, sad to say.


That's exactly what the UG does. I think the "fast" sampling rate available is about 1/second, which isn't very fast for a gauge. The consumption figures that it shows you are approximations, every single one of them. No car that I know of (except possibly some very specialized and $$$$ test cars) measures fuel flow into and out of the fuel tank. So you get estimates of fuel used based on the amount of time the injector is open, and (if you're lucky) some notion of the fuel pressure versus the manifold pressure. And maybe a fudge factor for the amount of time it takes to open and close the injector itself. Still an approximation.

And the engine management system doesn't know the exact amount of torque or HP that it is producing at any time. There aren't any sensors that provide that information. It may be able to guess, using the vehicle speed and a time counter, and some guess as to the current weight of the car and current traction available and air density and so on, but that's an approximation of an approximation.

So again, you've got graphs of things approximated by multiplying approximations, plotted against things approximated from guesses.

It would take a decent amount of work to get results, and the results would be about a half-step better than random noise.

IMHO. (Well, not that humble!)

-soD

Our test benches run expensive (accurate) dynos that are calibrated very accurately, and v expensive fuel measurement systems, along with fuel conditioning to hold temperatures accurately (40degC +/-0.5degC). This last point is critical as variation in temperature changes the fuel density and therefore the measured flow into the system. The air is supplied at controlled conditions (25DegC and 40% relative humidity, from memory).

Even with that, and with our dynos at sea level we then correct for atmospheric conditions.

Personally, I don't think you could get anything meaningful from an OBD measurement - sorry!

what you could get is instantaneous mpg.

which is meaningful.

I agree completely with soD and JMCC. The only way I know of determining power with any reasonable degree of accuracy is on a high quality load dyno.
Accurate fuel flow measuring instruments are expensive and as was mentioned, fuel density has a major effect on the outcome. I calculate my fuel flow based on accumulated injector on time over a set time period. The best I seem to be able to get is about +- 3% of actual averaged over a range of speeds and loads.

For back to back under controlled conditions maybe. Question was about generating a BSFC map and the errors in the fuel calculation in the ECU make this too inaccurate to be useable (in my opinion).

Reason for inaccuracy: fuel flow is calculated as a function of rail pressure and injection duration for a nominal injector. Injector flow particularly will vary, and needle lift vs electrical pulse will also be inconsistent.

OBD is accurate, but not precise.

What are you going to do with BSFC once you know it on a chassis or engine dyno?

Probably try to figure out most optimal engine parameters to run the engine at.

But you would need to compare that to precise measurements of the aerodynamics of the car, which would require a lot of time in a wind tunnel with a moving floor.

Another option is to drive your car down the road with a MPGUino and learn where your car gets the best MPG.

Sure it won't be PRECISE, but it will be hella accurate.

An Alternative to the traditional Dyno wold be a sensor in the drive wheel or drive shaft that measures the real time torque and RPM in real world conditions.

The companies that I know of that will make such devices for you are in the $10,000 to $20,000 range ... that same amount of money would buy lots of dyno time... and such a device would be specific to that wheel and or shaft ... it could only be used on other vehicles using the same wheel or shaft.

Either way you still need accurate fuel flow rate measurements... measurements that are celebrated for the conditions properly ... things like temperature , pressure, etc.

The other thing I would add to the list not already included by others would be a calorimeter to test the energy content of the fuel being used ... Consistency in testing is important ... the energy content of that gram of fuel is a potential source of error if it is not accounted or tested for... a non-quantified 1% variation in fuel energy content would limit the entire systems accuracy to less than 1% accuracy... there are a variety of factors that can alter the energy content per gram of fuel by more than 1%.

Some Can bus OBD2 ports will supply 20-100+ samples per second. Some cars are lucky enough that there are open source tuning solutions to monitor and/or reflash the factory image/program. Aftermarket engine management systems such as the megasquirt build a VE table automatically from the wideband sensor and other constants such as the fuel injector size and pressure, as well as variables such as the current calculated injector on time ( duty cycle). Over time just driving the car the VE table is 'auto tuned' from a target A/F ratio.

Another aspect that could help is measuring the power output of the engine without a dyno. There are mobile phone apps as well as windows apps that will do a pretty accurate chart of your power output. The fudge factors are the aero and rolling resistance corrections. Tbhe way these work are for the mobile phone, it just records the sound of the motor, and uses the frequency of the sample + a clock to figure out the rate the engine is changing. The windows one works similar, except you can use the line in port on the side of the laptop with a simple 5 component circuit to protect your laptop from voltage spikes / transients and take a direct ignition signal in as if it was an audio signal. The the computer, audio is just a data stream to deal with. StreetDyno is the program name that comes to mind off the top of my head. Avoid the apps that use an acceleraometer as the input as they are not that accurate. the output of the StreetDyno was within 5% of the numbers generated at the same time on a dynojet.

Very much so for feedback on tuning your driving style! And that is one of the major things that the UG, the SG, and the MPGuino all provide. The absolute numbers are not all that accurate, certainly not accurate enough to come up with a BSFC map, but they are very useful for driving and A/B/A testing and so on.


According to friends of mine who have used that setup, it only gets you close. There is still quite a bit of hand-tuning that needs to be done, either on a dyno or out on the road.


And weight, which you would have to measure to have more than a guess. (That includes the weight of fuel, driver, etc.)


I'm not saying you can't get useful numbers, and I'm not saying that you can't generate numbers for a BSFC map. I'm just saying that the final map you get won't be worth the effort you have to go through to get them.

-soD

Once again I agree with soD. I am a great fan of Megasquirt and of the Tuner Studio tuning software. While I'm sure that there are others who will disagree with me, I feel that the Auto Tune and even the VE Analyze features don't really work "as advertised" in all situations, but I think they can get you close. In my opinion, open loop tuning with lots of time and effort yields the best results.

I agree too.

With Wide band O2 sensor, I believe that auto tune can get you somewhere in range and from there manual tuning will make best out from system.

However when you would like to get best out from it, spark control needs to be set too and for my understanding that must be mapped manually and change to that often needs change in ve map too, so any automatic feature will just get one somewhere in range and best result will come with manually tune them all.