Long-Term Road Tests

A full tank is 20 lighted bars on fuel gauge. Each represents 0.368 gallon.

The city inspector has signed off on it, so our Phill home natural gas pump is officially up and operating. For the 1,395 miles the Civic GX has traveled since at-home refueling began, average fuel economy has jumped by a little more than 3 miles per gallon – an extra 21 miles per tank  as best I can figure it.

Warning: What follows is a lengthy discourse on how we are computing fuel economy with a system that doesn't tell us how many gallons are being pumped into the tank.

For those who don't care about the how,  the net result is that we're now averaging 29.38 miles per gallon in our compressed natural gas Civic GX after 13,562 miles, up from a little over 28 mpg at 10,000 miles. The boost comes mainly from the average of 32.34 mpg we're recorded over the 1,395 miles that we've been using the Phill.

If you do care, or are hoping to catch me in an error (not hard to do), feel free to read on 'til your eyes glaze over.

Physics explains the jump, but I'm not a physicist, so the best that I can do is pass on the explanation I received from Eric Rosenberg, assistant manager of American Honda's alternative fuels unit:

The Phill pushes more compressed natural gas into the car's tank than does a rapid-fill pump at a retail station;  so even though the fuel gauge reads full with either type of fill-up, there's more fuel, so more miles per gallon, in a tank filled by the Phill.

It's all about heat and pressure.

The Phill reads the pressure in the tank and fills until it hits 3,600 pounds per square inch.  Because it is a very slow fill, averaging the equivalent of just 0.4 gallons an hour, the Phill doesn't cause much of a heat build-up in the tank.

Retail pumps, however, shove the gas in at a rate of close to 2 gallons per minute, which causes a heat build-up which causes the internal tank pressure to rise which fools the pump into thinking the tank is full when it isn't. Additionally, the retail pumps merely pump til the fuel tank pressure is equal to the pressure in the pump's storage tank, and that's often well below 3,600 pounds

I've put 1,395 miles on the GX since the first fill-up with the Phill, and under a system co-developed with Honda's Rosenberg, figure that it's pumped the equivalent of 43.138 gallons, for a fuel economy average of 32.34 miles per gallon.

We averaged 29.05 mpg for the first 12,173 miles we put on the car and had been seeing fuel economy increase a little with each tank. 

The denser Phill fill explains the fuel economy increase, although, as Rosenberg admits, there's no easy way to get a really accurate read on gallon-equivalents delivered by the home pump.

Some gas utilities provide a separate meter for it, but here in Southern California, the Phill unit is hooked into the home meter, so it's hard to figure out exactly how many therms went into the car versus the hot water heater, clothes dryer, stove-top, gas fireplace and whole-house heating system. Best you can get is a pretty good estimate by comparing several years of gas bills on a month-to-month basis, and I've not received a new bill since I started using the home system.

Using the Civic's fuel gauge isn't any more accurate.

It has 20 bars that all light up when the tank is full, and then disappear one by one as the compressed natural gas in the tank is depleted. But with the Phill, at least with our Phill, the system shuts off and registers full when 19 of the bars are lighted.

Rosenberg says that he's received a mix of reports from GX users, with many others also reporting that their cars' tanks appear to be full when  that penultimate 19th bar lights up and that they never see the20th bar.

Amber light and no bars means you're into reserve of about one-half gallon.

We know that the GX's tank holds the equivalent of 8 gallons but that only about 7.5 gallons are usable. "Approximately" half a gallon always remains because by the time you get down to it, the pressure in the tank is too low to deliver it, Rosenberg says.

What's more, an undetermined amount of useable CNG – somewhere between 0.25 and 0.75 gallon – remains in reserve when the last bar disappears. I've driven 20 miles with a blank gauge, but the mileage depends on a slew of variables including ambient temperature and the vehicle's acceleration and braking pattern.

So the system we've come up with is a pretty straightforward averaging.

Because we've determined that our tank is full at 19 bars on the fuel gauge, with perhaps a gallon equally divided between the useable and unusable reserves, we've divided the remaining 7 gallons by 19 and assigned a value of 0.368 gallons to each bar. If nine bars have been depleted since the last fill-up and all 19 are lighted again when the Phill says the tank has been refilled, we multiply 0.368 by 9 and come up with a value of 3.312 gallons.

If the tank is run deep into that last half-gallon or so of reserve, we add it in and use 0.385 gallons per bar to calculate that fill-up.

This isn't exact – the Phill fills by reading tank pressure and the pressure is higher when the tank is half full than when it is empty.  That means that the last bars of a fill up contain a little less CNG than did the first ones. It all averages out in the tank, but means that by assigning a static value to each bar we get a reading that's slightly off when the tank is refilled before it is completely empty.

But, as Rosenberg says, "that's the best you can do."  Not even the alt fuels team at Honda, which has a vested interest in knowing as much about the GX, the Phill and CNG refueling as possible, has come up with a better method.

So until someone out there develops a gas flow meter that can be attached to the Phill's nozzle and gives an accurate gallons-equivalent reading, this is what we'll be doing.

I have emptied the tank as much as possible on two occasions since the first Phill fill (a Phill-up?) on February 28, logging 228.2 miles on the first tank and diving deep into the reserve, and  203.7 on the second., traveling only about a mile after the last bar on the fuel gauge disappeared. 

The rest of the time, I "plug in" the Phill's nozzle and turn it on when I park the car each evening, usually refilling somewhere between 8 and 11 bars overnight.  It can take 19 hours to refill an empty tank, and form 9 to 12 hours to do half a tank.  The last half fills more slowly than the first half because the pressure in the tank builds as it fills, and the flow of gas from the pump nozzle slows as it meets resistance.

If you can possibly want to know anything else about the Phill and how it works, the manual is available on-line and you are welcome to it.

If you've got a better way to compute mileage, or find a significant error in our methodology, we want to hear from you, so shoot me an e-mail.

John O'Dell, Senior Editor, Green Car Advisor @13,620 miles