How many tanks to determine gas mileage

Messages
1,979
Location
Houston
5 might be good enough for most depending on how much your style, driving habits, patterns, conditions, etc varies. I would say when your running average remains within +/- 1mpg you're pretty close.
 
Messages
3,931
Location
Somewhere in the US
There's a statistics class that deals with this exact subject - comparing 2 means. And the answer is 1) dependent on how much difference there is, 2) how much variability between readings of the same type there is, 3) how sure you want to be. But if you want a less technical way, I would start by recording your old readings first, then add the new values. That ought to tell you when you have enough data.
 
Messages
39,805
Location
Pottstown, PA
Why not just run perpetual mileage calculations? I've been doing this for years just to determine when stuff needs attention (tune, etc.). Any drift, one way or the other, should be clearly apparent if you do this.
 
Messages
345
Location
Northern California
More than one, thats for sure. I hate it when someone tells me they got some great MPG, better than mine but, under inquiry, it turns out it was only a one time check. Also, if you are like me, the driving changes from good to bad so it maybe hard to get more than a couple of tanks that have comparable driving. I also do mine every time I fill and record it in a little book for each car. To make sure, a person probably should reverse the change and see if the milage goes back down. Some think I am a skeptic, but I'm not sure.
 

Shaman

Thread starter
Messages
2,438
Location
Frankfort, Kentucky
Well, here is the situation Bought the car. Mileage was dropping from 7.5 to 6.75 at .25 intervals. Picked up FP and have gotten 7.25, 8.2, 7.7, and 8.5. The 8.5 was on a half tank, as gas dropped what would be 10 cents a gallon here. I also switched from AC constantly to none, back to intermittently. I will be keeping track of the mileage, but the excel file I was keeping it on got deleted, so I started a new one with the mileage that I remembered. I may also start storing it on a free PDA that I got, that I can take with me and input data as soon as I get it. The season changed about 6 Thursdays ago, and hasn’t been the same since, so the mileage I was getting, with the AC on, can’t compare really to the mileage I am getting with FP, and the AC off.
 
Messages
39,805
Location
Pottstown, PA
Don't you hate it when all those variables get in the way of figuring stuff out?? My wife is a good test platform for validation. She's driven the same way to work for about 20 years. She hits the same traffic every day. She's got no agenda ..and could care less if she squeezes out another .00003 mpg out of anything she drives. She's not going to indulge me by using a light foot ...nor sabotage me by using a heavy one. AC use is as needed and is never spared. She does a 15-20 second warm up before jetting away.
 
Messages
43,656
Location
'Stralia
I run a rolling spreadsheet, and the variance is amazing tank to tank (or even three tanks to three tanks). I've got a fuel card, and have used the same bowser, at the same station, and got 15% tank to tank. Only averages that I do are over OCIs. Wouldn't know how to factor a second (well technically third) variable with fuel additives. my 1990 4Runner when it was the family car was interesting. Track the mileage during an oil change, and it would take a (three tank) downturn at around 12,000 to 13,000km. Change the oil and it came back.
 
Messages
43,656
Location
'Stralia
this is the spreadsheet results for the current SUV. Note the difference tank to tank.  - the x axis is tank fill number. the y axis is kilometers per litre.
 

Kestas

Staff member
Messages
13,945
Location
The Motor City
Shannow - very good data! Now I'm itching to have you try 5W-20 to see what the real world data is on this oil's effect on gas mileage.
 
Messages
2
Location
Newton, IL
I have tracked my mileage since purchase of 97 Chevy K1500, not to mention oil changes, repairs, tires, fluid additions, etc. BIG Mileage increases were noticed at oil & plug changes. Then when I started logging where I was buying gas another big change when I kept buying at one gas station.
 
Messages
1,533
Location
Ephraim
quote:
Originally posted by Shaman: Well, here is the situation Bought the car. Mileage was dropping from 7.5 to 6.75 at .25 intervals. Picked up FP and have gotten 7.25, 8.2, 7.7, and 8.5. The 8.5 was on a half tank, as gas dropped what would be 10 cents a gallon here. I also switched from AC constantly to none, back to intermittently. I will be keeping track of the mileage, but the excel file I was keeping it on got deleted, so I started a new one with the mileage that I remembered. I may also start storing it on a free PDA that I got, that I can take with me and input data as soon as I get it. The season changed about 6 Thursdays ago, and hasn’t been the same since, so the mileage I was getting, with the AC on, can’t compare really to the mileage I am getting with FP, and the AC off.
so far so good looks like its paying for itself [Cheers!]
 

Shaman

Thread starter
Messages
2,438
Location
Frankfort, Kentucky
It's just above paying for itself, but without the use of AC, I can't constribute it to the FP. I may stop using FP for a few tanks, check mileage, try acetone for a few tanks, then nothing, then back to FP. By that time it will be time for the Jan oil change. Also need to check the tire pressure too. Need to find a place that can air them up here.
 
Messages
9,778
Location
Central Coast, Calif.
I'm taking a statistics class right now. the technical answer to your question is, according to the Central Limit Theorem, the "sample size has to be sufficiently large" for it to accurate. that number is generally accepted to be 30 samples. [freaknout] I think you could get an indication after 5 tankfulls but to be more accurate and account for driving habits, season changes, fuel changes, etc you would need to watch your mileage for a much longer time.
 

Shaman

Thread starter
Messages
2,438
Location
Frankfort, Kentucky
Shannow, I started an excel graph with my KMPL, and I am also noting different variances that would clearly cause a difference in testing, like the use of fuel additives, and the use of AC. Will also be noting if I went on any long trips.
 
Messages
43,656
Location
'Stralia
Shaman, it's interesting how the lines wander. The 4 peaks involve trips to the NSW central coast, sea-level (we're about 800m here), with a fair bit of cruising at 120km/hr, ait con on. The 4 (about 3/4 through the M1 oil change) that are right on the OCI average are 2600km in 56 hours, all backroads long distance stuff.
 
Top