1engineer
Moderator
I totally understand. The links you quoted are the proper way to test and that's not gonna happen here. Do you know what a PID loop is? I'm just trying to advance the real data up the curve towards the target a bit. Before you shoot holes in it lol hear my logic:Figures 5 & 6 would disagree with the assumption that time and distance will even everything out
http://www.epa.gov/smartway/forpartners/documents/sae-reports/evaluating-fuel-economy.pdf
Those error bands are quite large. Thousands of miles still won't account for the slew of variations that will creep into this type of measurement. Tolerances are a pain! Here's one of many sites that just scratches the surface of why MPG typically improves during warming months:
http://www.topspeedracer.com/better-gas-mileage-in-summer-than-winter.html
Are these variables controlled by time and distance in the proposed test? I don't believe so. There are other things that can be measured to show the functionality or lack thereof of the eco block and we would be better served monitoring those.
One guy goes 50 miles and reports his instant mpg reading.
Or one guy goes a few tankfuls and reports his mpg.
The second guy would be more accurate but still a long way from being worth anything right?
So far that is what we have - a few guys going a few hundred miles and reporting their findings. It's mostly meaningless.
Now flounder has volunteered to test one. He drives several thousand miles a month over the same route. He is not buying one so he has no reason to fudge. He actually keeps records.
Now this is just one guy driving. His actual mpg is meaningless because of so many variables compared to the next guy right? Right but his percent increase, decrease or no change would be applicable to othere. Given, this is one guy in one situation but in my mind this would be better than the guy going 50 miles and reporting his instant mpg.
In addition do a trailing average on his actual results will have the desired effect of smoothing out the weekly variables that are invariably present in day to day operation. In this case the longer the test runs the more accurate the results will be... for one data point. I volunteered to document this so everyone would at least know this one test had no bias.