gasoline prices

[cdn_in_bj]

The gauge lowered faster from a half tank compared to driving on a full tank at the same speed.

Traditional gas gauges are very inaccurate - what is indicated as half-full is usually much less.

If you don't believe me, you can try a little test. Wait until your tank is almost empty (warning light comes on), then fill with half the capacity (as stated in your owner's manual) of your tank. Then, check your gas gauge. It'll most likely indicate that the tank is much more than 1/2 full (I bet closer to 3/4 full).

[Rincewind]

I was a taxi driver for five years and can tell you categorically that you should keep your tank as close to empty as possible. I have tested both situations over a distance of over 10,000 miles and there is a big difference between the mpg of a near empty tank and that of a full tank. I would typically put in just one or two gallons each day and run around most of the time with the warning light on.

The gauge in the car is not linear. It is designed to be less sensitive when the tank if full and move more rapidly when the tank is getting empty. The middle point in the gauge is not half a tank. When your gauge is in the middle you have less than half a tank left. So, bhchao, on your two part trip you use more than half a tank of the first leg and less than half a tank of the second leg which explains why you got different mileages for each.

[cdn_in_bj]

I have tested both situations over a distance of over 10,000 miles and there is a big difference between the mpg of a near empty tank and that of a full tank. I would typically put in just one or two gallons each day and run around most of the time with the warning light on.

Which can be attributed to the weight savings of running on a near-empty tank.

There are some downsides to running with a near-empty tank, such as shortened fuel-pump lifespan and allowing condensation to form in the tank.

[gougou]

Isn't there a possibility that the gauge isn't exact?

EDIT: OK, guess that's been said already - didn't see there was another page...

[mr.stinky]

weight savings? the car probably weighs in around 2000 pounds. gas tank likely

holds 15 gallons, somewhere around 100 pounds when full. so the difference between

a full tank and half tank around 50 pounds. this will have no significant effect on gas

mileage, and no way to accurately measure it given your odometer and fuel gauge.

you'd first have to remove all the other factors.......speed, type of driving (city/highway),

temperature, wind speed and direction, tire pressure, road surface, blend of gasoline

and impurities, etc.

and don't forget this is a taxi. pick up one fare and you've just added 180 pounds, a tad

bit more than the weight differential of the gasoline.

[cdn_in_bj]

weight savings? the car probably weighs in around 2000 pounds. gas tank likely

holds 15 gallons, somewhere around 100 pounds when full. so the difference between

a full tank and half tank around 50 pounds. this will have no significant effect on gas

mileage, and no way to accurately measure it given your odometer and fuel gauge.

Rincewind did say that he ran with the tank near-empty, not half-full. So we can say that the differences he was dealing with (between a full and near-empty tank) would be greater than 50 lbs. I'm not saying that this difference in weight would totally account for the improved fuel economy that he observed, but it's the only direct factor that I can think of.

you'd first have to remove all the other factors.......speed, type of driving (city/highway), temperature, wind speed and direction, tire pressure, road surface, blend of gasoline and impurities, etc.

Agreed.

and don't forget this is a taxi. pick up one fare and you've just added 180 pounds, a tad bit more than the weight differential of the gasoline.

In his example, the fare is a constant (it's not like he was driving around all day with a full tank of gas and not picking up any customers).

[mr.stinky]

"it's not like he was driving around all day with a full tank of gas and not picking up any customers."

you don't think this would have more of an impact than the weight of fuel? what percentage

of time does he have no passengers, or one, or three, or carry several suitcases to the

airport? this factor is too variable, will more than offset any change in weight of fuel.

oops, late for class.............

[flameproof]

Weight has an impact on car performance and therefore fuel consumption.

If you drive a taxi an even bigger impact will have luggage in an open boot due to wind draft.

A simple flag on the outside of the car will increase petrol consumption by about 10%

Aircon cost's at least 0.3 L/100Km

[adrianlondon]

Maybe the liquid sloshing around in the tank also reduces fuel economy, especially for stop-start driving.

[Rincewind]

mr.stinky and co, you are all making the error of putting micro issues into a macro problem. The measurement was done over distances of 10,000 miles. The only condition changed was the amount of fuel in the tank. In both sections of the test the car was identical, the number of customers transported was near enough identical, the manor of driving was identical, the style of road and quantity of traffic was identical. It was a pure test on whither a full tank or an empty tank was more efficient.

The difference in weight is significant. Particularly over the distances a taxi travels. Over the course of a year it would be enough to pay for a weeks holiday in Spain or similar. Something well worth having. For the ordinary punter out there, the saving would be smaller and the inconvenience higher, but waste not want, not as they say.

and no way to accurately measure it given your odometer and fuel gauge.

The measurement does not use the fuel gauge. Any measure done with that instrument will be so full of errors as to be meaningless. The measurement is not over one tank, it's over 10,000 miles and 60+ refills. The receipts for the fuel show precisely how much fuel was added each time, accurate to +/-0.005 litres. Over many refills, this rounding error will cancel out giving your may hundreds of litres accurate to a very high degree. This error will further cancel as both tests are subject to the same rounding error so giving a fair comparison. The odometer on a taxi is accurate and subject to regular testing by the licence board. So the distance of the test is precise.

[flameproof]

The receipts for the fuel show precisely how much fuel was added each time, accurate to +/-0.005 litres.

The amount you buy is correct (hopefully), however, they amount put in can vary a lot, easily 5%. Depending on how much time you are willing to spend to really fill it (specially with diesel, with does foaming during filling). It also depends if you get the air out completely or not, which is a gamble.

In short: the method you use is not scientifically accurate.

What I do to measure my consumption:

I drive the tank near empty, that 700-800Km for me, I get it down to usually less then 1 liter, often 0.0 (according to the board computer - that I am able to drive more then 5Km with 0.0 shows that the board computer is also not accurate)

Then I fill the tank and calculate the L*100/Km - that give me the Liter/100Km. Interestingly the board computer is mostly a bit lower then what I calculate. Seems it's friendly towards the manufacturer.

The amount that goes into the tank varies also a lot whether the car stays level flat, or if stays slightly downhill, that way I get more in.

[Rincewind]

The amount you buy is correct (hopefully), however, they amount put in can vary a lot, easily 5%. Depending on how much time you are willing to spend to really fill it (specially with diesel, with does foaming during filling). It also depends if you get the air out completely or not, which is a gamble.

In short: the method you use is not scientifically accurate.

No sorry you are wrong. I'm not filling the tank to the same level every time. It doesn't not matter if my tank was filled to 79 litres in it today and 79.5 tomorrow and 78 the day after. What matters is that I put in 10 litres today, 9.95 litres tomorrow and 16.26 litres the day after. I know exactly how much petrol was put into the tank over the course of of the test.

Any measurement done over the course of a single tank of fuel will be so full of errors as to become meaningless. You don't know for sure how much fuel is in the system, you don't know exactly when you get back down to the same level again. You may have an error of as much as a +/-0.5 gallons over a measure of 15 gallons. That's a error of +/-3.25% when measuring something that might only vary by +/- 2.5%. So you don't know if what you did really changed your milage or was just a random error factor. Even if you got the error factor down to a little as 1 litre, it would still make the results of the test vague.

In my test the quantity of fuel involved as hundreds of litres, the error of measurement of each fill up cancel each other out so the only potential for error as the initial starting value and the final finishing value, many months later. This would likely be similar to your error, except I measured form full to full rather than from empty to empty. But look how the error factors workout. Say for example, of the full test I used 900 litres and the error was the same s you, +/ -0.5 gallons (these are example numbers not form the real test). Then my error factor would be just +/- 0.25% so the result would have significance.

In addition, a single tank of fuel is a short period of time. You will find it difficult to make several tests with identical conditions. A single event might radically affect the result. Where as my test was conducted over many thousands of miles. Individual freak events will have less significance and will occurred reasonably equally to each measurement.

[flameproof]

Brief me again what this was about.

I have also a record for my car after about 36,000Km, but didn't bother yet to calculate the average since after every filling I am in a similar range.

If that was about that a full tank is more efficient then there is another error. Each filling will be different by a certain amount of fuel. That absolute different is similar no matter if the tank was empty, or near full. Therefore, with a half full tank the error becomes larger since the percentage gets larger.

Since 2 days or so we have HK$16.96/L for regular. That's US$ 8.23/Gallon ... and whiny USAians complain about their $4-5

...and I got the feeling it will never be as cheap as today....