Variation in Fuel Used Monitoring

[jimh]

On my boat there are two systems that can measure fuel used:

--an Evinrude ICON Pro RPM gauge, and

--a Lowrance HDS-8 display in conjunction with a Lowrance EP-85R Data Storage Device.

Both of these fuel usage systems are configured to get their data about fuel flow rate from the propulsion engine's electronic controller. The engine is an E-TEC 225-HP; its EMM sends data about fuel flow rate to a NMEA-2000 network. Both fuel usage systems get this data about fuel flow rate, which is then integrated over time to calculate a volume of fuel used.

In the past I have noticed that the two fuel usage systems do not track precisely. On a cruise this summer (2016), I kept records of the fuel used data produced by these two separate systems, and can now offer some observations about a comparison of them. The data was collected over a week, during which about 70-gallons of fuel was consumed. (Since the two systems each calculated their own fuel usage, for simplicity I have just used the figure from one of them as the total fuel consumed; the HDS showed 69.6-gallons. I did not note the ICON Pro gauge's total. The initial fuel remaining was 58.8 on the HDS. A total of 33.5-gallons was added, for 92.3-gallons; at trip-end 22.0 was shown remaining. This suggests that total fuel used was 92.3 - 22.0 = 70.3. This number may be off slightly because of errors in the fuel pump dispensing the fuel and in conversion of liters to gallons by the author. This may account for the 0.7-gallons difference in total fuel used calculations. )

At the beginning of the cruise, the fuel tank level was at about 7/8-FULL, and the fuel remaining data on the two systems was slightly out of agreement. This lack of agreement was due to the two systems having been tracking the fuel remaining value for some time without having been arbitrarily set to the same volume, as would occur when the fuel tank was filled to its capacity. But even though the two values did not match, a comparison can be made by just tracking the difference over the week as about 70-gallons of fuel was consumed. Here are the data:

FUEL REMAINING
At cruise start:
HDS = 58.8
ICON = 55.4
Difference = 3.4 gallons more on HDS

At cruise end:
HDS = 22.0
ICON = 19.9
Difference = 2.1-gallons more on HDS

To be clear, the data shows that the HDS system calculated that less fuel was used than the ICON system. That is, the ICON system calculated slightly more total fuel was used. The difference in the calculation of fuel used between the two systems was a total of (3.4 - 2.1 or) 1.3-gallons during a period in which 69.6-gallons were consumed. This difference expressed as a percentage is then 1.3/69.6 or 1.9-percent.

To account for the difference in fuel used shown, one must realize these values are calculated values that are obtained by monitoring the data from the engine about rate of fuel flow and integrating that data over time to find a volume. Differences in the method used by the two systems could account for the difference in the calculated value of fuel used. For example, one system may sample the fuel flow rate more often than the other. If one system samples the flow rate every five seconds while the other samples every 30-seconds, there is a possibility that the system with the finer granularity of sampling might be more accurate. (I have no idea how often the two systems actually sample the fuel flow rate. I suspect the engine reports its data probably at a rather high frequency, say once per second or higher.) There is a reasonable basis to expect that the two systems may not agree precisely due to some difference in their methods. This might even be something like the number of decimal places used in the math being done, or, as I speculate, in the time intervals between samples. In any case, the data observed shows that the two systems calculated the fuel used value with an agreement of about 2-percent or better.

While it would be wonderful if, as boaters, we could know the exact value of fuel remaining to a precision of better than 2-percent, such a greater precision would not be particularly useful. In a tank with 77-gallon capacity (as in my boat) a variation of 2-percent is only a 1.5-gallons. There is no practical situation in which knowing the value of fuel remaining with an error of less than 1.5-gallons would be critical. The usual practices of fuel management in vessels is to allot about one-third of the total tank capacity for reserve, which in this instance would be about 25-gallons.

For the two systems of calculating fuel remaining to agree to several decimal places would also be wonderful, as that would indicate they were both reading the same data and calculating using the same methods. Such close agreement between the two systems would suggest that any error would come from the data source, the engine itself. But having the two systems agree with a precision of about 2-percent seems reasonably assuring that the methods are valid.

On my boat there is a third method of calculating fuel remaining: a dial pointer gauge linked to a float in the fuel tank. This data is rather coarse data, and you cannot calculate the fuel remaining with anywhere close to 2-percent accuracy. The tank level indicator probably can tell you the fuel remaining within a range of plus-or-minus five gallons. The ICON and HDS fuel remaining data is typically twice as accurate.

In the past there have been reports of significant discrepancies in the value of fuel used calculated by systems like these, particularly with the Lowrance EP-85R and HDS (or other Navico multi-function display). The results obtained during this cruise seem to be very consistent between the two systems, and also to be in good agreement with the (coarser) indication of fuel tank level from a mechanical float gauge. To judge the absolute accuracy of the two fuel used systems is difficult, because we do not have a third method which can be used as an arbitrator and which has recognized absolute accuracy.

The usual approach to fuel used measurement is to measure the volume of fuel added that is needed to re-fill the fuel tank to a known level. This method has two significant problems: measurement of volume of fuel added is taken from a retailer's fuel dispensing pump; observation of the re-fill level is usually not made directly, and generally comes from the dispensing pump shutting off automatically. In both realms there is a concern that all retail fuel dispensing pumps have precisely accurate and identical calibrations, that their automatic shut off always occurs precisely at the same boat fuel tank level, and that the boat's orientation during the fueling is always the same.

In MIchigan the required calibration accuracy allows for a tolerance of plus-or-minus 6-cubic-inches per five gallons. Five gallons is 1,155-cubic-inches, so the accuracy is thus 6/1155 = 0.5-percent. However, in a three month period in 2004 in which 953 retail fuel pumps were checked for accuracy, 15-percent or 115 pump were found to be out of tolerance. The pump which was most out of calibration overstated the fuel delivered by almost 100-percent, i.e., showing 5-gallons when only 2.5-was dispensed.

When two different pumps are used in the refilling of the the tank, even if both are accurate to the required 0.5-percent, they could be off in different directions. This means the total error could be 1.0-percent. Also, there is no limit on a pump understating the volume delivered, as that is not directly regulated, other than by the retailer's self interest. Given that substantial fines are often imposed for over-stating the volume delivered, retailers might chose to err in the other direction, resulting in a retail fuel dispensing pump to tend to understate the volume delivered.

The trigger point for the automatic shut-off of fuel from the pump can also vary from pump to pump, and no doubt on the same pump if other factors change, such as the temperature, the orientation of the filler nozzle, and so on.

The boat's orientation when fueling is also a factor. Static trim on the boat may be inconsistent if the fueling occurs when the boat is in the water. If the fueling occurs when the boat is on its trailer, the grade at the pump may be different, resulting in the boat's orientation being different.

All of these factors tend to make precise measurement of fuel used by re-filling of a fuel tank to a particular level to be somewhat inconsistent and open to variation.

[alloyboy]

Seems strange that two devices would do simple math and not come up with the same answer.

I actually prefer a fuel flow measurement type of system (Floscan for instance) that actually measures physical fuel flow as to opposed to a calculated fuel flow system. Many variables in the calculated system which can't and don't stay consistently steady the entire time. Injectors open and close at different rates with differing voltages, fuel pressure regulators have tolerances which allow fuel pressures to vary somewhat, fuel injector orifices become contaminated decreasing fuel flow, injectors can stick open increasing fuel flow, etc., all of which can cause actual fuel flow to differ from calculated fuel flow. And then as you note, perhaps the data being sampled by an MFD from the engine ECU/EMM/ECM is taken at different rates which could affect the calculations.

Having said all of this, a fuel management system (calculated or measured) should be secondary to a fuel quantity indication system whereby quantity available is taken from a fuel quantity probe and displayed on an independent gauge. IMO. Many love fuel management and think it is the superior system. While generally accurate, it has its idiosyncrasies.

[jimh]

Re reliance on a direct fuel tank level indicating system in preference to an indirect and calculated method: on my boat there is a directly indicating fuel tank level gauge. If there is any disagreement about fuel tank level among the three gauges, I trust the directly indicating gauge the most.

Re measuring fuel flow rates with an in-the-fuel-hose sensor: the flow rate of many modern outboard engines at low engine speed is very low, perhaps as little as 0.2-GPH. An in-hose turbine sensor must be extremely sensitive to be able to accurately measure flow at rates that low. The typical flow sensor provided with inexpensive flow monitoring devices is not up to the task of accurately measuring fuel flow at these really small flow rates, and it will miss a lot of fuel flow time and volume.

Modern engines must meet EPA emissions standards, which means their engine controllers must deliver the proper amount of fuel to the cylinder. The burden of maintaining EPA compliance probably necessitates accurate fuel delivery to a degree that exceeds the requirements for accuracy of fuel management calculations for volume used. If two systems of indirect fuel volume calculation come up with totals that are a small percentage apart, the error is most likely in their internal method and not in the data they are getting.

[alloyboy]

I called and talked to Kevin at Floscan, [who is a] very nice man and very helpful. Of their four typical fuel flow sensors, he said [their model] 20B will provide reliable readings down to 0.2 to 0.3 gallons per hour. He indicated they have other sensors that will provide accurate results below these values if needed, and 0.2-GPH is just a dribble.

For an old dog like me, it is amazing what modern technology can measure. Unfortunately, we can now measure and see that our drinking water has one part per trillion of some undesirable substance, which psychologically scares us. The stuff might have always been in the water. Difference is we are now able to measure and see it.

I see that some E-TEC outboard engines are reported to be burning 0.08 gallon per hour. Yes, eight hundredths of a gallon. Only with modern electronics are we able to calculate such a low fuel flow rate, I suppose. But then only with modern electronics are we able to deliver fuel to a particular cylinder using an injector that can be opened and closed for just a few milliseconds.

I guess if a man were a belt and suspenders sort of a fellow he might use engine computer derived fuel flow data correlated against Floscan type derived fuel flow data correlated against fuel quantity data. Then some would drive themselves crazy when the numbers don't agree to the third decimal. One downside of the digital world I suppose. On another forum a poster was going crazy because at any particular engine RPM the voltage displayed on his MFD would vary by maybe a half volt over a period of time. He just knew that something was wrong with the electrical generation system.

[jimh]

Seems strange that two devices would do simple math and not come up with the same answer.

The FLO-SCAN turbine sensor is probably sending the same data: rate of flow. Exactly how it comes up with volume is not clear. It might work like a Taft Rail Log and count the turns of the sensor to get volume; or, it might sample the flow rate periodically to get volume.

When I had a in-hose flow sensor, the flow rate was over-reported at low engine speeds compared to the engine's own data. And the flow rate was under-reported at high engine speeds, again, compared to the engine's own data.

[porthole]

When I had a in-hose flow sensor, the flow rate was over-reported at low engine speeds compared to the engine's own data. And the flow rate was under-reported at high engine speeds, again, compared to the engine's own data.

I don't know if the Lowrance in-line hose fuel sensor is a turbine or impeller type, but I do know that I can't get it to work properly--or get Lowrance to offer any tech support.

[What are] the workings of the Lowrance fuel flow sensor?

I've gone as far as re-installing my Yamaha gauge so I had some type of fuel use reading, especially since adding the Lowrance fuel level sensor negate using the traditional analog gauge.