Potential feature and disagreement on presentation
I disagree with how Fuelly displays its MPG; this is really only an issue in the context of a feature I'd love to see on the site: categorizing fuel-ups by gas station
One of my favorite functions of Gas Cubby (iPhone app I have been using prior to joining Fuelly) is that I can assign a gas station to each fill-up, and then filter my fill-ups by that station and see my MPG from fill-ups from there.
When using this feature, I can get, over time, a picture of comparative MPG with different brands of gas. Costco gas is cheaper, but am I getting worse mileage there?
But that data is only valuable if I get a lot of it. Via crowdsourcing on Fuelly, though, we might get a much better picture much faster!
(I realize that there are challenges: how valid are the results? How consistent are the users? Is the population of drivers similar at Station A than at Station B?)
Given that context, then the MPG display on Fuelly is wrong. The site currently displays MPG for Fill-up X and associates it with Fill-up Y. The gas one gets is assigned to data from a different set of gas!
Let's say I pull into the gas station to fill up. The odometer reads 1000 miles, and I buy 10 gallons of gas. Let's say that's my first-ever fillup, so Fuelly (correctly) doesn't calculate the MPG.
So I drive for 300 miles and I fill up again, this time 9 gallons. That's MPG of 33 and change. That 33 mpg came off of the gas I bought in the first 10-gallon tank. That MPG belongs to that gas. But on Fuelly it is displayed with the SECOND tank, because that's the moment of purchase.
My contention is it should be listed with the FIRST tank, because that's the gas that yielded that mileage.
On Fuelly, currently the first tank always shows no MPG. This other way, the LAST tank will always show no MPG (until the NEXT tank).
If you're not taking data about the gas itself (i.e. what brand or station), that may not be of much consequence. But perhaps you're taking notes on this stuff manually. I want the MPG to be assigned to the purchase of its source gas, and not the purchase of the NEXT gas.
I hope I've made what I'm asking clear. Hope you find this input helpful! Regardless, this is data that I'm really excited is crowdsourced. It really did inform my opinion when shopping for cars!
I know about a site where I can choose (from a drop-down box) the brand of gas, and I actually fill in this info (which leads to the same offset issue lkalliance mentioned), but there are no other features there, it's just for curiosity.
Other than that, I've filled many different brands and batches, and it's very rare to see any measurable difference between them, except for a very few exceptions when there was a 'something's wrong' feeling anyway. The worst of those was a deliberate experiment of mine, when I tried a self-concocted E33 mix...
Once upon a time I attempted to see if I could measure a difference in MPG between "pure" gas and E-10. The car was a '94 Pontiac Sunbird with a 2.0 liter engine and automatic transmission. The driving was pretty standard involving a 30 mile commute to work. Fill-up was usually at the same station. I was unable to measure a consistent difference. Any such difference seemed to be "buried in the noise" of different mileages caused by weather changes and such.
In theory, E-10 contains about 3% less energy per unit volume than "pure" gas. At the approximate 30 MPG I was getting with the Sunbird, that would have been about 1 MPG. Tank-to-tank variation usually exceeded that.
The problem with your approach is, what if you fill up your tank from halfway with brand X. Now at the next fillup after this (from empty) you label this tank as "Brand X" but it is actually half brand X and half of the brand before it. It is no less foolproof than what we are doing now, and it is certainly easy to figure out when you are using spreadsheets just by shifting a column up by one row. And certainly if you are doing any sort of meaningful statistical analysis, you must be dumping your data out to excel...
I too tried for years to find a difference between different brands of gas, and indeed still log the brand of fuel at every fill up out of habit. I could never find any evidence for one that is statistically significant (which is why I fill up at Costco almost exclusively now). Ethanol is a different story -- but it's almost impossible to find pure gas at a reasonable price anymore.
I've never found someone who could back up the claim that two brands give different MPG with real data. It is always speculation. Though I am always open minded to data that says otherwise.
At least in my area of Nebraska, there are a lot of stations which sell gas without ethanol, along with E-10. Typically the E-10 is about 10 cents per gallon less expensive and about 2 octane points higher. This gives me the choice of fuels. E-10 outsells "pure gas" by a large margin because most of us tend to buy using price as the only important criterion.
For the last year or so I have been trying to flag my Fuelly entries with "E-10" as a tag, when I am reasonably sure that is what I get. In many locations the pumps are not marked, so I leave the tag blank. I realize that when filling a partially-empty tank the result will be a mixture of the two fuels. However, after several fills (five is probably a good number) of the same fuel the result will be essentially unmixed.
There are a few things here that I think you're missing. First, as another poster pointed out, the switch back and forth from Ethanol blends in states and areas that allow it would definately skew your results, here in California the switch to and from "summer blend" makes a difference in our mileage even with all gas being mandated to be at least 10% ethanol for years now. Second, and a driver for me is the "cleaner" gasolines that don't clog fuel filters in my vehicles. This isn't as big a deal now as it was 10 + years ago, but I still tend to shy away from a couple of major brands after having to change fuel filters that were clogged after several tanks of "dirty" gasoline back in the 1980's. I also stay away from one brand that Chevrolet attribuited to clogging fuel injector poppets on my 1998 Chevy S-10 Blazer and all GM vehicles equipped with a 4.3 liter V-6 and V-8 engine applications. According to the Chevrolet service bulletin there was an additive in southern California gasolines that was required to meet AQMD requirements that was gumming up the poppets and causing them to stick, GM did a recall and warned against future use of So Cal Chevron gasoline. Finally, most independent gas stations and even some major brands will buy fuel from other refineries if there is a supply interruption. I've seen brand x trucks at brand y stations before and a family member who drove gas trucks years ago has told me that a fair portion of the time even though he drove for Shell, his orders would have him fill his truck at another refinery to make his deliveries.
I've not found much of a difference, either, but that depends on how you define "much". There are two stations at which I primarily fill, and I've found over the course of a year of data that one has given me 1 mpg more. But what I hadn't thought of was what you pointed out about any given tank really being a mixture of the remnants of the previous tank (however much that is) and the new added gas. Darn it.
I guess all that's left is to just continue to scratch that obsessive-compulsive itch, lol.
We are working on our road map for Fuelly and being able to link your fill up to a gas station is something we would like to see as well. This is wonderful feedback and we'll keep the issues you mentioned in mind when designing a solution. We know it won't be perfect but the data could be pretty interesting, especially on a large set of data (as well as by State, Country, etc).
We don't have any ETA on when something like this might be incorporated. We are currently focused on getting the iOS and Android apps published. Ideally we could build this into the apps so they know your long/latt and can automatically figure out which station you are at when you are filling up.
The Fuelly/road map thing will only work if people use their smart phones to enter data at the point of fuelling. For those like me who enter data later, when they get home, the data will be misleading at best. I usually (but not always) fuel my motorcycles at the station 3.1 miles from home. I usually fuel my pickup at stations either 20 or 30 miles away, depending on where I happen to be (has to do with credit card reading pumps). This isn't likely to be very helpful.