Date Odometer Gas price Gas vol Gas total Distance True MPG Avg. MPH Display MPG Error%
11/26/2013 73057 3.539 13.161 46.58 249 18.9 19 20.1 6.2%
11/27/2013 73382 3.699 12.937 47.85 325 25.1 67 26.5 5.5%
12/1/2013 73638 3.979 11.487 45.71 256 22.3 28 23.4 5.0%

I keep track of (most of) my gas use. The initial intention was simply to see how much I was spending on gas. The other reason is that I did not trust trip computers and I wanted to compute my true fuel economy. I think manufacturers have an incentive to give an overly optimistic MPG figure. Or maybe their algorithm of calculating MPG is not good enough and ends up overly optimistic, either intentionally or unintentionally. I realized I can verify whether the trip computer is telling the truth with information from the gas receipt and the trip computer. Later I discovered other useful metrics in the trip computer that would allow me to get some more insight into my car. There are missing data here and there, especially earlier in the log, but over the past year I managed to do fairly good record keeping, and now I have a good amount of data that are worth presenting. So, here they are.

Is Audi trip computer (TC) honest with me?

Whenever a friend talks about his gas mileage I would question him whether he read that off the TC or did the calculation himself. The answer is mostly the former, and then we get into endless arguments. I don’t trust TCs. People like to see high MPG numbers. It won’t surprise me if car makers cater to such need. Speedometers overstate your actual speed (albeit for good reason), so why should I trust MPG readings?

The idea is to calculate gas mileage myself and compare it with what the TC tell me. The method I use is simple and common. Every time I fuel up, I record the reading on the odometer, D, and how much gas I purchased to fill up the tank, V. The difference between the current (D) and the previous (D’) odometer reading is the distance I traveled using that V gallon of gas. My gas mileage for the past D miles is then (D-D’)/V miles/gallon. Easy enough.

The Audi TC has two “computers”. Computer 1 automatically resets itself every time you start  the car, unless it has been less than two hours since the car is shut off, in which case Computer 1 will not be cleared automatically. Computer 2 can only be reset manually.  Upon each fill-up I will clear the “average consumption” on Computer 2 (press and hold the button at the bottom of the wiper stalk) and not touch it until the next refueling. The reading I get upon the next refueling is then what the TC thinks is the gas mileage since my last gas station visit. For some manufacturers this is even simpler because the TC automatically resets itself after each refueling, but Audi’s system is straightforward enough and easy to use as well.

After I have the results of both I calculate the percent error of the TC number, assuming my method is perfect.

Here are the error percentages of the last 30 entries. The average is 3.6%, with a standard deviation of 5.4%.

Histogram of Error%
Histogram of Error%

Surprisingly, my TC is not that much off.

My method of calculating gas mileage is not perfect, but I try to avoid as many variations as possible. I go to the same gas station and use the same pump whenever possible. I try to park at the same location relative to the pump. This is because my gas station is built on an unevenly sloped hill. Even an inch off means my car will be tilted at a different angle, which might affect when the pump decides the tank is full and stops fueling. There are also variations inherent to the pump that cannot be accounted for.

Fuel Economy and Speed

The next thing I am interested in is the relationship between MPG and speed. It is well-known that cars (excluding some hybrids) use less gas on the highway. But I want to know a detailed function of fuel economy vs. speed. The Audi TC records average speed, which comes in handy. The relationship is shown in the graph below.

Fuel Economy vs Average Speed for 2007 Audi A4 2.0T Quattro

Unfortunately in most cases I either drove mostly in the city or mostly on the highway so I have very few data where average speed is between 30 and 50. I included a simple fitted line generated by Excel. I picked second-order polynomial because visually it seems a better fit. An exponential curve actually generates the least R-squared, but there is one caveat – it trends upward at the far right. This will not make sense because driving at 100 MPH is apparently less efficient than driving at 60 MPH.

So for this car the most economical speed is somewhere around 60 MPH, far below what people normally drive at on the highway. The car will be in 6th gear and engine speed just below 2000 RPM. This is one of the few moments when I wish I had a newer car that has a wide ratio CVT or 8-speed Automatic.

Again, this is not a very scientific study. There are way too many variables that I did not control. Weather (temperature), what is loaded in the car, type of fuel (Some say BP gets better gas mileage while Shell gives you more power), route/destination (going somewhere higher elevated demands more gas), etc., all have an impact on fuel economy. But you get the general picture.

Amount of Fuel Used between Gas Station Visits

Where is the needle on your fuel gauge when you go to the gas station? In other words, what is the minimum fuel level you are comfortable with before you have to fill it up? Apparently it’s personal preference, and one that I believe has something to do with personality, risk aversion, sense of security, etc. Maybe some psychologist would be interested in this. My own impression is that I usually refuel when the needle is between the 1/4 mark and the red (1/8) mark. Now that I have the data, let’s give it a concrete measure.

The fuel tank has a capacity of 16.6 gallons (17 if non-Quattro). That minus the amount of gas purchased is the amount of gas left when I go to the gas station. I would fill up the tank before going on a road trip regardless of how much gas is in the tank, so I excluded entries where less than 6 gallons of gas were purchased in order to better reflect my habit during day to day driving.

On average, I refuel when the tank is 24% full, which echos my impression.

I have heard from experienced drivers/driving instructors that, even when the fuel gauge shows zero, you are still good for another ten, twenty miles which translate into roughly a gallon of gas. It makes sense to me if car manufacturers understate how much is left in the tank so that people don’t run out of gas. Pessimism doesn’t hurt in this case. However, according to my not-so-scientific study, fuel gauges are honest.

I will update this post as more data come in.