Excellent point, SoopahMan. Gas guzzlers might speed up the tests, but why compromise when you can take the axis and stretch it a bit to see fine distinctions?

 

This is a side issue, but high-sulphur gas (if they still sell it) is not good for the life of your catalytic converter - who wants to replace one?

 

I just finished testing Citgo, nothing remarkable, but new information is changing the next series of tests. The Motor Trend review of the Lexus 600h had a small footnote recommending Texaco premium and the Wiki article on octane suggests that premium fuels may have more energy per unit volume. This suggests that I need to survey premium grades too.

I'll start with Texaco and Shell. Then I'll do a discount brand, Costco, both regular and premium. This will let us compare and contrast the grades within two brands and a generic.

As for sulfur, my last hill climb followed one vehicle that really stank. I almost wanted to catch them and ask were they bought their gas. But short of having some sort of field test, what protection would an individual have against 'sour' gas?

Bob Wilson

 

Interesting thread, especially since Shell and Phillips are our only two local choices for gas.

We've been buying Phillips exclusively for our non-hybrid cars since Shell bought out our local BP dealer. The reason was that the non-hybrid cars ran really badly on the Shell gas. We pretty much had to buy one grade higher to get the cars to run without knocking. Not so with the Phillips.

But ... perhaps it is worth giving Shell a try with our new Prius. We're still on the tank that the dealer bought for us, though ironically enough that was Shell. I guess we will be experimenting with both brands.

 

After hearing all this favorable information about Shell, I was amused today when I saw this headline:
Shell gets Surprise Profit Boost: http://news.bbc.co.uk/2/hi/business/6618195.stm

 

GASOLINE BRAND INTERMEDIATE REPORT
(V01.00)

SUMMARY:

• Of the brand gasolines tested, 87 octane Shell delivers more energy per gram, ~12%, than 87 octane Conoco, BP, Texaco, Citgo and 93 octane Shell and Texaco.
• Costco 87 octane gas has similar performance to 87 octane Shell.

BACKGROUND:
Recently, someone posted that the brand of gasoline was important for mileage and that using "Shell" gave them better miles per gallon than other brands. To test this, I used my 2003 Prius, NHW11, equipped with a Graham Davies mini-scanner to perform a hill climb test with a sampling of Huntsville AL brand gasolines. Within experimental limits, Shell 87 out performed all other brands and even two 93 octane samples.

The Graham scanner reports MG1 and MG2 torque and rpm, which can be used to calculate the ICE torque, rpm and power output in watts:The Graham scanner data was recorded to a laptop during a 55 mph climb of an altitude change of 525 ft. / 160 m. for a length of about 1.2 miles, about a ~6% grade hill called "Brindley Mountain." To ensure gasoline sample quality, the car was driven until fuel exhaustion and then on electric mode up to a mile to a safe area to put in one gallon of the next test gasoline. Then four more gallons of the test sample were added and at the next opportunity, a hill climb test.

All hill climb data above 2,400 ICE rpm, ~22,000 W / ~30 hp, were used with the fuel injector timing to calculate the specific fuel consumption, Watt_seconds per fuel gram. However the tests spanned temperatures from 60-85F / 15-30C. The first and last Shell 87 octane tests were performed at the temperature extremes. The Costco 87 test was from data captured last September, likely to be at least 85F. The other brands were tested mostly in the 60-70F range.

FUTURE:
The Dept. of Energy and a racing product report that the energy content of petroleum products is a function of specific gravity. A hydrometer was bought to measure the gasoline density but unfortunately the wrong scale was ordered. The unit will be exchanged but that will take sometime. During this wait, I am driving on a full tank of Shell 87.

The final report will include gasoline density for these brands but not these samples. Regression testing will be used to make sure at least two of the next gasoline samples show this density correlation and a wide selection of local gasoline brands will be density tested including non-brand, discount gasolines.

Bob Wilson

ps. The typical gasoline energy rating is ~46MJ/kg. Using 16,000 W_s/g and a 33% ICE efficiency leads to ~48MJ/kg. However, we are in a region of decreasing ICE efficiency (the gasoline didn't change!) and it is nice to see the correlation.

 

Be aware Cosco buys their gas from whatever company gives them the best deal, so the gas can change from time to time. Usually done on longer term contracts, so it should be stable for months at a time though. Ditto for grocery stores (Superstore, Safeway, and Saveon Foods here in Canada) as well as Sears and others.

I had always heard premium contains less energy than regular, due to the additives used to boost the octane rating taking up space normally occupied by molecules that provide lots of energy. So your test bears that out anyway.

Does your fuel now always contain some ethanol? That is what I thought I read in a Fed. Govt. mandate. Ethanol replacing MTBE and Techron.

I use Husky/Mohawk fuel here in Canada. It contains ethanol. Therefore a little less energy. I use it because it burns cleaner. Check the throttle body on your car (remove rubber hose on intake and look for soot). Husky/Mohawk use the ethanol to boost the octane rating, whereas Esso added tolulene. Tolulene burns badly and produces soot in the engine. Xylene is also used sometimes. It's toxic, but I suppose gasoline is anyway.

Great graphing job, BTW. Are you using a Canview?

 

Hi David,
I'm looking forward to getting a proper scale hydrometer. If the energy content turns out to be proportional to density, we'll have a quick test.

I have a Graham miniscanner and record the data to a laptop.

Bob Wilson

 

Bob- Great effort, however, very disappointing, and provides zero useful data.

If I read correctly, your test sample was:
"525 feet of 6% grade at 55 MPH"

This is 6.5 seconds of driving Bob. This is way too small of a sample to mean ANYTHING. A 5 MPH breeze could account for your variances. Pure chance, coincidence, could account for all your variances. Atmspheric pressure (weather) could account for all your variances.

Sadly, the "noise" in your data is larger than the data you are trying to measure.

Bob--> to have something meaningful, please test a sample of at least 10 gallons or some "test route" over 100 miles long, with each gasoline.
Then, as a rule of thumb, you need to test each gasoline 3 times.

If you do that, I'm 99.99% sure all your gasolines ( of same octane ) will match very, very closely. Probably they will match more closely than the noise in your 100 mile tests.

People in labs in major univerisities have done these kinds of tests on the bench-top with hardly any of the variables you have to deal with, and in these controled conditions, show just the opposite... generic gasoline has the same btu per gallons as leading brands.

Is is always possible to find a "fluke" such as ground water in a tank at a particular station that would reduce performance of the gasline from that particular station, but not effect the gasoline brand-wide.

BTW, water is much heavier than gasoline, and a tiny amount will skew your density ratings. In this case, the LEAST dense gasoline is best.

 

One quick comment for those not familiar with the scientific method. We perform an experiment, publish the results and discuss the findings. It is often in the discussion that all parties gain new insights or testable hypothesis for follow-up experiments. This discussion is a sharing of opinions about the facts and data and nothing personal.

I see the problem and corrected it to read:Each fuel sample set consists of ~85 samples covering a little over 1 second each. But let's take a look at each factor.

"A 5 MPH breeze" would be a valid concern if vehicle drag were part of the measurement. However, the energy measurements were taken at the shaft input to the transaxle given by the ICE torque and rpm and fuel consumption given by the ICE rpm and injector timing. A head or tail wind would have moved the average watt-seconds but the injector timing would (and did) follow.

Now I agree that any experiment should have enough runs to eliminate the possibility of "pure chance." Looking at the data, the unusual grouping of all brands except Shell 87 suggests that "pure chance" was not at play. Even the two 93 octane samples share a similar pattern including a slight upward trend at the higher power settings. Then we find the second Shell 87 sample looks closer to, but not exactly, the same as the first sample.

Now given the change in density as function of temperature and the ~60F temperature of the first Shell 87 run and the ~85F temperature of the second Shell 87 run, one would expect the 'warmer' gas to show up as less energy. Sure enough, that is the relationship between the first and second runs. The other runs were in a narrower temperature band, 60-70F. Fortunately, this is a testable hypothesis and worthy of a couple of test runs.

The predicted temperature changes on Saturday and Sunday are 61-85F. I have a full tank of the same gas used for the second run. I have no problem with doing a pre-dawn and a mid-afternoon tests at the minimum and maximum temperatures. I would expect to see the apparent energy density to be lower in the afternoon than the morning. This will be fun.

Sometimes we have to deal with noisy data and that is what the 'trendline' function in excel does. There are several trendline functions and I'm using the "power" function.

I appreciate the suggestion and have started a full, 11.3 gallons of Shell 87 to see if a full tank using my regular MPG calculation methodology shows a similar improvement in MPG. My usual practice is to calculate MPG using pump and trip meter mileage. However, it will take a while to get any meaningful results using this methodology.

In the meanwhile, others may elect to use their MFD mileage displays and say run a 1/4 tank sample to dry of different gasoline samples. This would give meaningful results in a fairly short time period, say one week per sample.

Or 'you' might perform the proposed test. You are aware of the experimental challenges and presumably have access to different brands of gasoline in your area. Follow your proposed protocol and post the results.

You may find this Oak Ridge National Laboratory report interesting:

http://bioenergy.ornl.gov/papers/misc/energy_conv.html

This is the original quote that gave me a clue about using a hydrometer to measure gasoline density to get an a quicker reading of the energy content. Once I exchange the hydrometer for one with the right scale for gasoline, I'll be able to do a quicker survey of gasoline brands and do a comparative study of the least and densest brands. If the hydrometer test works as expected, I'll be able to test gasoline at the pump before filling the tank.

BTW, I would be very interested in reading those major university reports. Do you have any URLs to such reports?

I've tried Google but not had a whole lot of luck beyond oil testing standards, military testing fuel testing protocols and racing products that advocate using density to measure the energy content of petroleum products.

Sounds like an excellent test when I get the hydrometer. I'll measure a sample of gasoline. Then I'll take out the hydrometer and add a teaspoon of water and shake the column to mix the two. After letting the water settle out, I'll use the hydrometer to measure the new gasoline density making sure it does not touch the excess water in the bottom of the column. This will give us quantitative numbers. Great idea!

What this means is if 'water in the gas' is a problem, two relative density measurements of a gasoline sample will tell if the gas is already water saturated. If both readings come out the same, the gasoline is water saturated. If there is a difference, the gasoline is relatively water free.

One word of caution! Testing gasoline with lab glassware involves handling a terribly flammable liquid whose vapor is explosive. This must be done with careful attention to the safety hazards. This means being away from any ignition source and anything of worth that might be damaged in a fire. Spills must be avoided and cleaned up properly and a gasoline fire extinguisher be available.

If it turns out that gasoline density testing is effective, I will investigate safer density testing systems. For now, I would strongly urge folks without such training avoid this potentially hazardous methodology until it can be made safer.

Bob Wilson