I did this for my own knowledge and figured it might be useful or at least interesting to the members of this forum, so I decided to post it. Since roughly the beginning of the year, I've been keeping track of my Blazer's fuel consumption, mostly because I was noticing that the on-board computer reading was inaccurate with regard to fuel usage (it'd say I used ~15 gallons, but the tank would be topped off after ~14), but also because I'm a statistician by profession and enjoy data analysis. This isn't the end of my data-logging, but I finally had enough data (35 entries) to actually do something with it.

The data I recorded was pretty basic... the number of gallons I filled up with (based on the fuel pump, which is regulated and must be within certain specifications), the odometer reading, and the average MPH and average MPG readings that the Blazer's computer gives me (both are reset at every fill-up). I also keep general track of the type of driving it was (mostly city, mostly highway, or mixed).

Here's the most basic analysis:
Overall MPH - mean = 42, standard deviation = 10.33, minimum = 20.40, median = 43, maximum = 59.7
Overall Actual MPG* - mean = 16.403, standard deviation = 1.787, minimum = 12.049, median = 16.629, maximum = 19.820

* Actual MPG is calculated by (miles traveled)/(actual gallons used), and is not the same as the Blazer's estimated average MPG. I'll address the difference later on.

As one might expect, fuel mileage is approximately normally distributed:


For those not familiar, this is a Normal QQ Plot, comparing the data to a normal distribution with mean = 42 and standard deviation 10.33. The closer the points are to the straight line, the more normally distributed the data is. If all points are between the curved lines (95% confidence interval), it's pretty much normal:


One point of interest is the impact of speed on MPG. Obviously, in general, the higher the gear you're able to maintain for longer, and the lower the RPM in that gear, the better the fuel economy. The government frequently tells us that speeds above 55 MPH decrease fuel economy due to the engine struggling to overcome aerodynamic drag... but that's generally bullsh!t. They want it to be ~55 MPH so that we obey speed limits on highways in cities, but in reality, modern cars are aerodynamic enough, geared well enough, and tuned to overcome this. I get the best fuel efficiency in my Blazer with the cruise control set to 65-70 MPH (the instantaneous MPG readout will sit around 17-18 MPG on level roads at these speeds). Unfortunately, due to traffic, ramps, construction, etc, I was never able to get my overall average MPH above 59.7, but regardless. Here's a basic scatterplot of how speed impacts fuel mileage:



If you look really closely, you'll see that a line which would best fit this data wouldn't be perfectly straight - it starts to curve down. This demonstrates that, at some point, the extra drag would start to decrease fuel efficiency... as I said before, I suspect it'd happen above 70 MPH, but I haven't been able to maintain an average speed that high for long enough to verify.

The next step is to cluster the data into driving types. Here's the gist of it:

City Driving
Average MPH = 26.49, Minimum MPH = 20.40, Median MPH = 26.3, Maximum MPH = 32.2
Average MPG = 13.98, Minimum MGP = 12.049, Median MPG = 13.941, Maximum MPG = 15.866

Mixed Driving
Average MPH = 42.295, Minimum MPH = 35, Median MPH = 42.6, Maximum MPH = 49.2
Average MPG = 16.640, Minimum MGP = 14.677, Median MPG = 16.615, Maximum MPG = 18.593

Highway Driving
Average MPH = 54.89, Minimum MPH = 50.7, Median MPH = 54.3, Maximum MPH = 59.7
Average MPG = 17.96, Minimum MGP = 16.737, Median MPG = 17.787, Maximum MPG = 19.82

I was actually quite impressed to see that I got nearly 20 MPG, once. My average speed on that trip was 58.5 MPH and I traveled 238 miles (one of my cross-Illinois trips that I make monthly, or so). Based on the Normal distribution of our data, I'll get more than 20 MPG only 2% of the time.

And here's a scatterplot of MPG vs Average MPH, clustered by driving type (including line of best fit):


(If you're wondering why the slope on the mixed driving line isn't as steep, it's because "mixed driving" has a lot of variance in it... it's never perfectly 50/50 city/highway, but is more one than the other, and thus the gas mileage is a little less dependent on speed and more on how many stop lights, traffic, etc.)

The last thing I wanted to figure out how I can estimate my actual mpg based on what the Blazer tells me. The computer was low by over 2 MPG, in one instance, so it can be pretty far off. When I fitted a regression line with 0 intercept, I found that the Blazer underestimates by approximately 8%. So:
Actual MPG = 1.08 * Blazer's Claimed MPG

If you don't have a calculator on hand, and you want to know a better estimate of your MPG, add 10% (move decimal left one place), then subtract 1% twice (move the decimal to the left twice and multiply by 2). So if your Blazer claims you're getting 17 mpg, you can determine it as follows:
17 + 1.7 - 0.17 - 0.17 = 18.7 - 0.34 = 18.36 mpg

As an example of how well it works in my case, the Blazer told me 17.1 mpg. Using the regression formula:
17.1 + 1.71 - 0.171 - 0.171 = 18.81 - 0.342 = 18.468 mpg
The actual mpg was 18.4792.

I don't know if this works as accurately for all Blazers... it may even be the case that mine is a rare occurrence, as far as inaccuracy goes. But I suspect it uses the computer's fuel usage reading (which, in my case, is off by frequently more than a gallon over a tank of gas) to compute MPG, so if you've noticed that your computer is inaccurate on fuel usage, it's probably also inaccurate on MPG.

Anyway, I hope you found it interesting (or informative if you're in the market for an S-10 Blazer and curious about gas mileage). Now off to General Tech to search as to whether or not the AC Compressor's clutch is replaceable.

 

pretty cool, and your findings are accurate, i get 13 city, 14/15 mixed and 17 highway, its scary how close your findings are to my mpgs. have you dertimined which octane gives better gas mileage?

 

I didn't do any octane tests, but now that you mention it, this actually began as a comparison of 10% ethanol vs. non-ethanol, and when I didn't see any significant difference over a few tanks of each (I'm fortunate enough to currently live in an area where non-ethanol fuel is still available), I just decided to continue for my own knowledge, having known about the inaccuracies of the on-board computer. Ethanol contains approximately 30% (or so) less energy than gasoline, so it was speculated that the fuel efficiency will drop significantly with ethanol content, but my analysis showed no significant difference. (I'm still anti-ethanol, though, due to my experiences with it destroying components in other cars I've owned.)

 

I recently purchaced a 1999 LS 4 door 4X4 about a month ago and milage is very lousy. A round trip from home to work and back, about 18 miles each way, uses right at quarter tank, even worse if I take the interstate. One thing I have noticed tho is that at times it will improve then suddenly reverts back to the poor milage. My driving habits aren't that bad. Can someone point me in the direction it would take to fix this problem.

 

tune up.

i was putting in about $20 every other day, did my plugs wires pcv fuel filter and a k&n air filter and now i dont quite have to fill up as often, oh and i can actually move now! the truck goes past 130kmh!

 

I love how I understood absolutely EVERYTHING you posted up there. I'm currently taking (not by choice admittedly) Experimental Methods in Biology working towards my BSc in Biomedical Biology and we've been doing some heavy duty ****.

When you do it by hand, it's so freakin time consuming.