MPG Computer - PeachParts Mercedes-Benz Forum

mach4 · #1 02-24-2014, 01:02 AM

Now that I'm done with my engine instrumentation project, it's time to start thinking about another project. In the thread for the instrumentation project there was some discussion about the possibility of adding some functionality to do miles per gallon. In researching the internet, there are lots of systems to display mpg using the ODBII interface on newer cars. These systems read the injector pulse width and frequency as the basis of fuel usage and tie into speed readings to get instantaneous mpg and distance readings to get mpg per tank. Obviously this can't work on early mechanical injector engines.

There are several hurdles to getting a system to work on older cars, especially mechanical diesel systems. The problems are

Finding a sensor that will read out fuel usage in the range of 2 - 20 liters per hour
Figuring out how to read net fuel used (diesels have a lot of return fuel going to the tank)
Dealing with the fact that diesels use fuel for cooling injectors and injector pump

I've found a diesel rated flow sensor that uses a Hall Effect sensor that can be read by an Arduino that reads 2-30 liter per hour. Some have suggested and even tried to read flows to and from the engine and and calculated the difference. The problem with this approach is that it's difficult enough to calibrate one sensor without the complexity of trying to calibrate two without any way to really do it effectively.

Years ago I installed a Zemco automotive computer in a Chevy Vega I re-powered with a Perkins 4.108 diesel. In this instance I used a float bowl from a Holley carb to create a closed loop system for fuel. The return fuel from the injectors went to the float bow, and I was accordingly able to read only the makeup fuel to the float using the included optical flow sensor. Now I was pretty ignorant of the cooling purpose of the return fuel and count myself lucky that I didn't fry something with that system. I thought about using a float bowl attached to a large surge tank, but wasn't confident I could get enough cooling to be viable.

In pondering these issues, I've come up with a potential solution. The following diagram describes a system to deal with the other two problems.

The solution I'm suggesting uses a surge tank/cooler to do the job. This would be a custom fabricated tank within a tank. The inner tank is the surge tank and the outer tank is for cooling. There are a couple of interesting features to this system. First, the fuel from the car's fuel tank is used to cool the fuel in the surge tank by surrounding it completely. A constant flow is provided by a low pressure electric fuel pump (think boost pump) that continually flows fuel from the main tank through the cooling jacket of the surge tank. As the engine uses fuel, that exact amount of fuel is replenished to the surge tank through a make-up line (the purple line in the diagram). Upstream of the makeup line is the flow sensor that reads the actual fuel used by the engine. The position of the T out of the pump helps to ensure that any air that happens to get to the pump is returned to the tank by being at a higher level from the flow sensor and having a higher flow rate.

I envision the surge/cooler tank, electric fuel pump and flow sensor all mounted together as an integral unit and tapping into the existing supply and return lines.

A couple of other thoughts. Obviously this kind of a system introduces points of failure that could strand the car - the electric fuel pump quitting, the flow sensor clogging, filter (not shown upstream of the flow sensor) clogging, etc. What I'm contemplating is putting 2 T valves in the system that would effectively completely eliminate the mpg gear in an emergency and return the system to 100% stock flow - flow from the tank would go directly to the lift pump and return flow would go directly back to the tank.

So I'm interested in thoughts and ideas from the PP brain trust abut the feasibility of this actually working. Is there anything I've missed in my thinking or research? ... or other ideas about how to build a real-time mpg system.

One more thought.... By using graphical displays like I've done with the instrumentation project, in addition to instantaneous and cumulative mpg numbers, that all systems provide, it would be possible to actually show a moving graph of mpg over time, which I think would be really interesting and informative. And from there, it would be a relatively small jump to include parallel graphs for altitude, speed, and even the boost, TIT and oil, trans and intake temps I've got in my other system, graphically showing the interaction and effects of all those factors.

Skid Row Joe · #2 02-24-2014, 02:13 AM

The onboard calculating, and real time mpg meter on my 06 E320 CDI, was a real eye-opener as per the when of certain mpg the car gets.

I'm really impressed with it's multiple capabilities, including an hourmeter and average mph readings.

I don't know how the car (06 CDI) does the data gathering, but I like it!

Hope you get the concept for the system you're considering turned into a reality.

Shortsguy1 · #3 02-24-2014, 08:57 AM

Mach4-
Since you started the engine instrumentation project, I have been thinking a lot about fuel economy and how to measure it. Ideally, it wouldn't require significant modification to the existing car. I think I will only implement mpg instrumentation on my cars if I can do it without re-piping any fuel lines. But how to achieve that has yet to be resolved.

I think you are right to avoid measuring the two flows and calculating fuel use by difference. As they say, measuring the difference between two large values to determine a small value is not very accurate. I would be very surprised if someone got this method to work well.

In a way, you are creating a second fuel tank. But your fuel tank, rather than having a fuel sender to measure capacity, has constant capacity and you measure flow in to maintain constant capacity. So in a way, if our fuel tank senders were perfect measurement tools, you could actually use that signal to indicate fuel usage. Obviously, the fuel height change in our tank is small and obfuscated by the waves, splashes, etc. as the car drives. Your approach is obviously better.

A few weeks ago, funola put a electric guitar pickup on an injector line to see what the signal looked like. For me, I suspect the holy grail of fuel economy measurements for our engines lies in measuring (indirectly) the injection flow rate at each injector. I suspect, but have no evidence, that the noise or vibration of the injector is a function of the flow rate through the injector. If I didn't have two young kids, and a falling apart house, this is something I would try to do.

The other idea I had was to use velocity, acceleration, altitude measurements by a GPS as a proxy for the power required of the engine. Some folks call it Road Load, and others call it Specific Power. But basically, you can get a rough estimate of the engine power with these measurements. Since the efficiency of our engines is not a constant value, you would need to incorporate one of the Brake Specific Fuel Consumptions maps which are available in the research literature. And then you would probably need to tell it how much fuel you used per tank, since nobody's engine is identical to the brand new MB data. While this method would only provide a proxy for fuel economy, it would probably be close enough for our purposes. Honestly, the OBDII values imply a level of precision which is not supported by how the measurements are made.

Please keep us posted. This is the engine instrumentation project which I am most likely to implement on my own cars, so thanks for all your hard work.

GregMN · #4 02-24-2014, 09:54 AM

Find the specifications/recommendations for maximum fuel temperature that will not damage the IP.

Loop your fuel return back to the fuel source line. Measure the fuel temp. and see if it goes out of spec. If so, install a liquid to air heat exchanger and see if that keeps it in spec. If so, you can now measure the flow from the tank into the fuel loop.

mach4 · #5 02-24-2014, 10:11 AM

Quote:

Originally Posted by GregMN

Find the specifications/recommendations for maximum fuel temperature that will not damage the IP.

I've been unable to find definitive information on that so far. So my design mimics the operation of an open loop system that is on our cars maintaining the fuel intake temps at the temp of the fuel in the tank.

I'm getting some interesting data off my instrumentation project that may be applicable. For example yesterday we went for a 100 mile run in the mountains and with an ambient temperature in the 70s the temp on my Arduino peaked at about 100, which is a good proxy for underhood temps at cruise.

Tmadia · #6 02-24-2014, 04:00 PM

I'm not sure where the return line is, but you could T the return line into the fuel line prior to the pump and then you could put your transducer on the line coming out of the tank. Nice and cool back there and the only fuel coming out, not going back in.

mach4 · #7 02-24-2014, 04:46 PM

Quote:

Originally Posted by Tmadia

I'm not sure where the return line is, but you could T the return line into the fuel line prior to the pump and then you could put your transducer on the line coming out of the tank. Nice and cool back there and the only fuel coming out, not going back in.

Anyone that's done a diesel purge knows that the volume of return fuel is not trivial. The concern is that having that a closed loop without any outside cooling as you suggest, would exceed the design parameters of the IP or injectors - especially in stop and go traffic in Phoenix in the summer, for example. Maybe it wouldn't but to my knowledge there is no empirical evidence of either what the design criteria of the IP cooling is or the temp that the fuel would get to in a closed loop system. I'm not willing to risk an IP or the chance of getting stranded on I-10 between El Paso and Van Horne just to experiment to find out. Without that data, my design is an attempt to replicate the cooling profile of the stock system in a closed loop system.

Tmadia · #8 02-24-2014, 07:33 PM

I know several people retrofitting older cars to fuel injection have done this to avoid running a return line to the gas tank and haven't had any problems.

But I understand your concern.

GregMN · #9 02-24-2014, 07:48 PM

Getting a digital read out of your mpg is going to be quite a challenge.

How about an analog reading of your relative fuel use. A gauge needle that reacts to changes in net fuel flow.

A flow meter in the source and return lines.
The gauge starts at zero.
The fuel source flow meter turns the needle one direction.
The fuel return flow meter turns the face of the gauge in the same direction.
When there is a difference in the two flow rates the needle will move relative to the face of the gauge.

oldiesel · #10 02-24-2014, 09:39 PM

Pretty common on large yachts. Called flo-scan

GregMN · #11 02-24-2014, 09:48 PM

FloScan Instrument Co. Inc.

mach4 · #12 02-25-2014, 05:12 PM

Quote:

Originally Posted by oldiesel

Pretty common on large yachts. Called flo-scan

Flo-scan is probably well beyond my budget.

The one I've been looking at is from Futurlec

This is a 2-30 liter/hour sensor and costs $13. They reportedly do pretty well at higher flows (such as normal cruise) but accuracy declines at low flow rates (such as idling).

mach4 · #13 02-25-2014, 05:31 PM

As I tend to do on these kinds of projects, I ponder solutions well outside the box and often come up with something that can work. My current line of thinking with regard to accurately calculating makeup fuel focuses on an adaptation of the method that the ODBII type systems use. Those systems measure and total the time that an injector is open. If you know the flow rate of the injector and the duration it is open, you have fuel usage. Instead of measuring fuel flow with an impeller and a hall sensor, or a ball in a race and an optical sensor, I'm wondering about the feasibility of using a very high flow injector (100#/hour range) tied to a pressure sensor in the surge tank. Under this scenario, the Arduino monitor the pressure and would turn on an injector in the surge tank whenever the pressure was at a low limit, say 3#. Since injectors are designed to turn on and off quickly you can shoot for near-zero hysteresis and therefore get to decent instantaneous fuel usage. The Arduino would measure the time that the injector is on and get fuel used based on the time. There are some difficulties with this setup that would arise in cold temperatures based on the change in viscosity of the fuel.

A similar, and maybe better option would be to use a low pressure, positive (and fixed volume) displacement pump of some sort that would enable just counting "clicks".

Just knocking around ideas that might lead to a solution down the road.