It is called a vacuum gauge and was done back in the 50's.

The higher vacuum you maintain, in theory, the better your mileage.  When
you press the accelerator your vacuum drops and that means you are using
more gas.

i believe a simple vacuum gauge hooked up to a port on the intake
manifold might do the trick, or at least be worth playing around with.
the less the vacuum, the better the mileage?

Driving the car in way that'll maximize on ecnomy requires a standard
tranny. Otherwise, you're at the mercey of the automatic.

The posted city consumption for my car is 7.8L/100km, but if I put my mind
to it, I can get 6.5L/100km. However, I'm not able to shave anything off the
highway mileage which is 5.8L/100.

Pars
98 Civic Hatch

This could be a fun project. I would start with a circuit like the one
below. Point A hooks to the switched side of one of the injectors, point
B to a high impedance voltmeter. Power supply to the 74HC14 should be a
regulated +5V supply, I am old fashioned and like the LM317 adjustable
voltage regulator for that purpose. The 180k / 100k resistors do several
things:

1. Voltage divider that converts the 14V / 0V swing at the input to a 5V
/ 0V swing for the 74HC14
2. Ensures that the circuit is a high impedance load on the ECU.
3. Provides high enough impedance at the input to the 74HC14 that
voltage spikes are effectively clamped by the protection diodes in the
74HC14.

The three inverters (with Schmitt trigger inputs) shape the potentially
ugly signal at the injector into a nice square wave with a 0 - 5V swing.
The 74HC14s are nice in that if they are supplied with 5V, the high
output voltage at moderate loads is very close to 5V.

The 100 k resistor and 10 uF cap form a low pass filter with -3dB point
around 0.16 Hz. This is a difficult trade-off. At idle you will still
see quite a bit of wiggle on your meter, and the response time is on the
order of 7 second. If you want faster response, you will get more wiggle
at idle. You could add more poles to the filter to get a better
compromise between response time and reading stability.

I would imagine that the duty cycle on the injectors is quite low, so
you might have to add an op-amp for gain between point B and your meter.
That will also allow you to add a gain adjust for calibration as well as
allow you to implement a more sofisticated filter.

Remember to ground the input of the three unused inverters.

                        74HC14
     180k          |\      |\      |\       100k
A ---\/\/\/---------| >0----| >o----| >o----\/\/\/--------- B
            |      |/      |/      |/                |
            |                                      ===== 10uF
            |---\/\/\/---                            |
                 100k   |                            |
                        V                            V

Now, if you like microcontrollers, you could hook the output of the last
inverter to the timer / counter input of your favorite controller (If
you still want to use the analog low-pass filter you probably would want
to use the second inverter rather than the third for your
microcontroller). Then you could time the pulses and calculate the duty
cycle that way. This will allow you to adjust your filtering to the
engine RPM so that you get a rock steady output regardless of RPM and at
the same time get quick response time on your meter.

A further experiment would be to use one resistive divider and one
inverter for each injector. Then use four diodes to OR the output of the
4 inverters together, and follow it with two more inverters. Then you
would get less wiggle on the meter at low RPM, or you could bump up the
cutoff frequency on the RC filter for quicker response. The
possibilities are endless...

Note: My 1:42 am engineering output is prone to errors. Use at your own
risk!

Old add-on gas mileage meters measured flow in the fuel line.
With some FI, you would also have to measure the return to the tank.
Then measure odometer, and do the math.