At the risk of being perceived as OCD, I thought that this would make for an interesting thread for you all.

I have been accumulating data on the RS over 100,000 miles. Why did I do this? When I bought the car, I didn’t know much about engines, and I thought by keeping a log of everything I would be able to detect when something was failing. I ended up recording every single tank of gas and every trip odometer reading over 100,000 miles (once I started, I felt compelled to keep doing it.) I also recorded when certain mods were done. With this information, it’s possible to do some observational analysis and simple statistical tests.

Methodology: 99RS manual, original owner, no aero modifications (no body kit and no spoiler.) Each fuel up was done by filling tank until the pump shut off automatically (never topped off.) 87 octane, California formula. The number of gallons added approximates the amount of gallons used since the last fillup, and some variation is to be expected with that method. The mileage since last fillup can then be estimated. Mods were not tested independently, as each one was added in succession.

Analysis: The data was first cleaned up by removing bad data points (occasional times when I did not fill the tank up completely so mileage could not be estimated). The data was coded to the presence or lack of mod. Since the data is from my daily driving and not some rigorously designed experiment, the analysis and outcome isn’t truly solid. So what, this is 2gnt, I used some t-tests. This is what happened…

Results:
Here is the overall mileage distribution over 100k



Mean mpg= 29.7, median=29.58.
Lowest MPG seen = 22.2, 22.5, 23.1
Highest MPG seen = 36.8, 36.4, 36.1
Avg miles to the tank = 355
EDIT: Most miles on one tank= 473
Avg gallons on a refuel = 12
Most gallons I’ve put in at once = 14.7

The lower MPG figures are likely attributed to stop-&-go traffic/racing/angry mood. The high MPG figures are accurate. These were times when I had almost 100% freeway driving, like when I went camping out in the desert or my commutes up to NorCal. And even then, my car was usually loaded with stuff/clothes/equipment/girls and going 85. With that in mind, I honestly believe my 2g could get 40 MPG freeway if I drove the speedlimit, did not have so much weight in cargo, and dumped the rich-running ECU for a MS running an optimal AFR.

Now, for the mods. Here is a timeline of sorts, when I put on certain mods and what the mileage was over time.



Now look at what happens around tank 40. See the big decrease in mileage almost overnight? (I put redlines around it to make the trend more obvious.) I laughed when I saw that, because I remember that actually happening. Like I said before, I didn’t know much about cars when I bought the RS and since I wanted to keep the car a long time I drove it like a grandma. Never went above 4k RPM, always drove conservative. Then one day I said enough - and started redlining it everyday and did my best to keep other drivers fearful of me. This was about the time I found 2gnt.com and realized how much you could abuse the 420a. I think that the blue line represents my encounter with you all.

Now for the mods. I ran tests on each mod before-and-after install. (There are issues with doing the analysis like that, but I won’t go into it.) The only modification that showed a significant gain (p<0.001) was the Magnaflow cat-back. There was a mean increase in mileage of 2.28 MPG.

Since the mods were added in succession, I tried to only compare a mod’s before and after results, when no other mods were being added. This reduced my sample size for some mods, so I lacked good power to test some things. For example, the Modern Performance throttle body showed a gain of 1.3 MPG, but this could have simply been due to chance.

The AEM V2 intake showed a decrease of 0.8 MPG, that approached statistical significance (p=0.063). There were only so many observations between other mods, however, and so this might have shown true significance if I waited longer before installing other mods. The decrease may be my fault, b/c I started to really like the sound of WOT after the intake install.

What’s interesting, is that the Unorthodox pulley and cam gears (set to stock settings) showed a significant decrease of 0.9 MPG (p=0.018). These were installed at the 60K timing belt change. It’s counterintuitive, but there may be a confounding factor at work here – namely the O2 sensor is starting to age.

I also ran regression analysis. The only significant factor was the magnaflow (beta coef= +3.1, p=0.002), and “time” (beta= - 0.019, p=0.0010). R-square 12.16.

Translation: A cat-back has been the only significant contributor to better mileage. All these mods contribute very little to predicting my mileage (explains only 12% of the variation), as mileage is mostly due to my driving behavior. Also, my mileage has been decreasing over time, as my MPG goes down 0.02 every time I fill up a new tank (still have the original O2 sensor.)

See guys, statistics isn't so bad

2gnt: '99 RS-T, killed by a toyota, pending rebuild...
Daily: Volt
Daily #2: '99 EVG ebike- STOLEN by PEDRO