Data on effects of weather
 

I've been keeping a daily log of my trips to & from work for about a month now. I don't have a Scangauge or anything like that, so I'm just using the factory installed trip mpg. I know it reads ~10% high, but I trust that it's at least directionally correct. Anyway, I've heard people talking about how temperature and other conditions affect mileage so I thought I would look at my data. I look at the records from a local weather station (from Welcome to Weather Underground : Weather Underground--not to be confused with the political organization) and record the temperature, pressure, humidity, & wind speed/direction. I thought others might be interested in the data.

FYI, I'm running a intercooled turbodiesel (with a variable-geometry turbo), so I am NOT implying that these results are generalizable. My commute is about 40 miles each way.

I've been changing (improving) my driving habits all along, so I realize this isn't exactly a controlled test. Also, the weather has changed here in Indiana in the last month from Spring to Summer. So I guess what I'm saying is that I realize this isn't the cleanest data set but I thought others would find it interesting.

I've separated my data into morning & evening runs to account for the inherent differences (too long to go into all the details right now).

Most of the trends make sense to me. FYI, I'm definig "Relative Wind Speed" as the windspeed realtive to the direction I'm heading. For example, if the wind speed is 10 mph out of the south and I'm headed south, the relative wind speed is -10mph.

I supposed the best thing the stats show is the the highest dependency (R2 value) is mpg vs day which means the strongest factor influencing my mpg is probably the nut behind the wheel becoming more well tuned! :D

Now this is interesting.

I see that there is significant and consistant change on your daily log. Where do your newer records show up on some of those other graphs?

It's all the same data, just plotted vs. different independent variables.

Like I said, it's not really a controlled test because all those factors are changing at the same time. Additionally, some of the factors are correllated. For example, the weather has been getting warmer, so the later dates tend to have higher temps.

I haven't done it yet, but there may be a way to "adjust" for different factors. For example, I could scale the number around according to date (so that the slope of mpg vs day = 0), and then re-plot. Then again, that would also inherently skew some of the othe factors which happend to change with the day.

IMO, the most interesting thing to look at are the R2 values. Plots with higher R2 values mean there is a stronger correlation. For example, you can see there is a stroger correleation between temperature and mpg than there is humidity and mpg. Therefore, I would conclude that temperature is more signifcant than humidity. Of course, the slope matters too...

Station pressure is un-corrected pressure, and along with temp are the two main factors in how much power it takes to push what air there is (thick or thin).

Those two are also the main factors in bullet drop at longer ranges and make a BIG difference in your elevation adjustments.

Wind also affects bullet elevation and sometimes to a greater degree depending on how strong it is.

Keep in mind that some factors influence the aero and the engine is opposite ways. For example, higher temps lead to lower engine efficiency (typically), but also lower aero drag (and presumably lower rolling resistance as well).

That's part of why I'm been kepping this log--to see what the overall effect is.

Have you measured the relative altitude of the home and work places?

The data suggests the work place may be a little higher than home.

Here is some tank data from our 2003 Prius:
http://hiwaay.net/~bzwilson/prius/pri_mpg_temp_110.jpg

and for the sane folks:
http://hiwaay.net/~bzwilson/prius/pri_mpg_temp_100.jpg

Here it is over a longer interval:
http://hiwaay.net/~bzwilson/prius/pri_mpg_temp_010.jpg
We bought the car in October 2005 so there was a little learning curve.

Bob Wilson

In my part of Indiana, it's pretty much as flat as a board. Technically my work is about 80 ft lower in elevation (over 40 miles that's negligable IMO).

FYI, part of the reason my evening mpg is worse is that (for most of this log), I've stopped on the way home to work every evening on a new house we've bought. It's about halfway, and I stop for about 2 hours, and have to go into/out of the new neighborhood.

"Adjusted" Plots
 

Here's the plots again, however, this time I've "corrected" for the change over the date. I adjusted the mpg by the slope of the "Day" plot, i.e.

Morning "Adjusted" MPG = MPG - 0.4076*Day
Evening "Adjusted" MPG = MPG - 0.3812*Day

Interestingly enough, the trends look pretty similar.

R^2 is your sum error squared?

Correcting for your improvements helped make the graphs much less noisy.

I think the evening teperature slope should be greater.

The morning wind slope should be decreasing, that also brings up a potential problem with that graph, can you double check your data something doesn't look right.