 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
Tony's Guide to Fuel saving gadgets |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Drive-cycle economy and emissions measurement Elsewhere on this site you will see why fuel consumption measured on the road is not a good way to judge whether a particular product / device really improves economy or emissions. Basically, the problem is that economy is very strongly affected by factors such as traffic, driving style, weather, etc, so that any small change is lost in the "noise". To get round this problem, industry and governments have developed a series of standard tests whereby a vehicle can be measured under completely repeatable conditions, and different vehicles compared fairly to each other. These tests can be generally described as drive-cycle rolling-road tests, and are conducted as a matter of routine on all new car designs. To avoid the vagaries of the weather, and having to transport the sensitive emissions measurement equipment, the test is conducted indoors on a rolling road (also known as a "chassis dynamometer" or just "dyno"):
Many years ago economy was measured at a constant speed, but it was recognised that this did not provide a good representation of how a car is actually used. Hence various standard "drive cycles" were devised, of which the two main ones are the FTP75 (US) and the ECE+EUDC (Europe). Most countries around the world follow one of these cycles. Both cycles involve starting the engine from cold and driving at various different speeds and accelerations for between about 7 and 11 miles (10 - 15 km).
The vehicle is "driven" over this speed profile and the various emissions recorded. At the end of the test, the total amount of fuel used is determined through the "carbon balance" method (the mass of carbon emitted from the exhaust must equal the mass of carbon in the fuel consumed) and hence the overall fuel economy is calculated. (This is what MLM Watchdog refers to as an "EPA Certified Emission and Mileage Testing Station Test".) This method gives extremely repeatable results, typically within 2% from test to test, and so is very easily able to detect even small improvements produced by an add-on product or device. This is in stark contrast to any kind of on-road testing, which (even when carried out carefully) always gives much greater variability. (See appendix for some typical test data showing the relative variability of road and dyno measurements.) An additional benefit of measuring fuel economy in this way is that the toxic pollutants are measured at the same time. Most people would agree that a device that improved fuel consumption by 5% while increasing emissions of toxic pollutants (say) ten-fold should not be widely used or sold, because of the negative effect on public health. (Due to the requirement for extremely precise engine control to keep the catalyst working correctly, such a trade-off is a very real possibility - for example with air bleed devices) Equally, data for all resonably modern vehicles measured over this test cycle is widely available from the manufacturers. If a test vehicle before the fuel "saving" device is fitted shows results that are significantly worse than its "official" data, and it then improves to the "official" value once the device is fitted, you have to wonder if there was some significant problem with the test vehicle to begin with.
Many people say that these tests are flawed. It is true that the economy obtained by a particular driver making a particular type of journey will not, in general, be exactly the same as the "standard" figure. How could it be, given the huge variation in driving styles and journey types? For example, my current car returns anything from 25 to 40 mpg depending on what type of journey I am making. Does that mean the official figure of 33 mpg is 25% too optimistic, 21% too pessimistic, or about right? The main point of contention is usually that the standard drive cycles represent relatively gentle driving patterns (in terms of acceleration and top speed) and so anybody who drives "like their hair's on fire" will get significantly worse figures. Motoring journalists, who specifically seek out quiet, twisty roads for car testing, often find this effect - but for most people, most of the time, driving on congested roads, the standard drive cycle is not too far from reality. More "aggressive" cycles (such as the US06) have however been developed to minimise this effect, and results from these would be equally acceptable. (In January 2006 the US EPA announced that in future it would be including results from the "US06", among others, in the "sticker" mileage figures, for exactly the reasons described above.) The point is not that the official figure should exactly match what any individual will obtain, but that it gives a good indication of how two vehicles compare to each other, or the effect on a single vehicle of making a change (such as adding on a fuel "saving" device.) The test does after all cover pretty much all driving conditions - cold start, hot start (FTP75), slow driving, fast driving, cruising, acceleration, etc, etc - so anything that fundamentally affects how the engine works should be expected to show up on this type of test. The test covers almost the entire engine operating range so it is hard to see how some add-on device could give very large benefits in the real world while miraculously giving no benefit at all under test conditions. In fact, the standard test is mostly biased to the operating conditions where the engine is at its least efficient, and so there is the greatest possible chance for a fuel "saving" product to have an effect. Another criticism is the use of the "carbon balance" system. The question is asked - reasonably enough - "why not simply measure the amount of fuel used?" The problem is that this is actually much more difficult than might be imagined. Most modern cars use a "returning" fuel system, where the fuel pump delivers a constant supply to the engine; a small fraction is injected, and the rest returned to the tank. So the amount used is the small difference between two large amounts, which is difficult to measure accurately. Another alternative would be to brim the tank, do the test, then see how much fuel is needed to fill it up again. Here the problem is the expansion of the remaining fuel - suppose a car has a 50 litre (about 12 gallon) fuel tank, and uses 2 litres on the test. The remaining 48 litres may expand or contract slightly due to temperature changes and so give a false reading. For example, a rise of 10 celcius gives around a 1% change in volume (48 to 48.5 litres) and so the measured consumption would fall from 2 to 1.5 litres (a 25% error!). (Of course, the same problem may occur with on-road testing.) The "carbon balance" system avoids both these problems. A specific claim sometimes made - usually by makers of fuel "saving" devices that have shown no benefit - is that the reduced fuel use due to their device is exactly balanced by the increased carbon produced as deposits are removed from the pistons, cylinder head, etc. Even if this were true, the effect should soon disappear over the course of a few miles' driving as the amount of carbon in the deposits is relatively small. Any well-designed test sequence should include enough running to eliminate this possibility - this could include running the vehicle on the road for a few hundred miles between drive-cycle tests. More implausibly still, some makers say that vibration is essential to make their device work, and this vibration does not occur on the rolling road. Clearly any device mounted on or near the engine will be significantly vibrated during rolling-road testing, and in fact the whole vehicle does shake quite considerably during the test (especially during acceleration and deceleration, as it moves on the tie-down straps.) But even if this effect really were real, why do the makers not simply fit some kind of vibrating device (say an electric motor with an out-of-balance wheel) to anything tested on a rolling road? A final attempt at mocking these types of test is to say, "How many people drive their cars on a dyno?" The implication is that there is something fundamentally different about driving a car on the road and driving it on a dyno. But how is the burning air/fuel mixture supposed to know the difference? Why should the fundamental physics of combustion be affected by what the wheels are attached to?
A common argument put forward by the makers of fuel "saving" products, who do not want to do this type of test, is that the cost is prohibitive. Figures in excess of quarter of a million US dollars are frequently quoted. Ignoring for a moment the fact that this is a drop in the ocean compared to the potential profits from a proven-effective product, such claims are completely untrue. When I worked at Cosworth Technology we did such tests as a matter of everyday routine (not specifically to test fuel-saving products, but as part of the engine development process), so I know what they cost. I was also consulted by Ecotek about their scientific testing, and still have the original quote for the work from their test house. Based on this, I know for a fact that a basic test program (two vehicles, two tests on each in "baseline", "modified" and "back to baseline" condition) should not cost more than £10,000 / US$20,000. Such test evidence would not represent incontrovertible proof, but would be far more comprehensive than I have ever seen offered by the sellers of a fuel "saving" product, and more than enough to defend against action by regulatory authorities. Even a much more detailed test program, with more vehicles and fuel types, is unlikely to cost more than ten times as much. So what's stopping these companies from doing the tests - other than a strong suspicion that they will prove the product doesn't work?
Many people ask why I am so positive about this type of drive-cycle test, but so sceptical about emissions tests of the sort typically used for annual inspection tests (like the Smog Check in the US, or the MoT in the UK). Aren't they basically the same thing? The answer is - no, not at all. Here are just a few of the important differences: Initial conditions: Emissions vary greatly according to the state of the test vehicle - temperature of oil, coolant, and (especially) the catalyst make a very large difference. Although vehicles are always in a "warmed up" condition for an inspection test, there can be huge differences depending on how the vehicle was driven immediately before the test. On a drive-cycle test like FTP75 or ECE+EUDC, the vehicle is "soaked" at a constant 25C for at least 12 hours before the test begins, to ensure consistent start conditions. Load and speed: The annual emissions test is usually only conducted at one, or perhaps two, load/speed conditions (typically, these are just idle and fast idle). So the engine is only run over a very small part of its operating "envelope", leading to results that may not reflect how it will behave in the "real world". The drive cycle test covers most operating conditions, including cold running, and so gives a more meaningful result. Emissions measured: Many annual inspection emissions tests completely miss out NOx (oxides of nitrogen), one of the most dangerous pollutants. Modestly reducing HC and CO, while greatly increasing NOx, is very easy to do - but very bad for the environment. It's essential to measure NOx to ensure this is not happening Mass versus concentration: The annual inspection test measures the concentration of pollutants (in % or ppm). But this does not tell you actually how much of each gas is being produced - if the engine uses twice as much fuel, and so produces twice as much exhaust, then the mass of pollutant gas will double even if the concentration stays the same! The drive cycle test actually measures the total mass of exhaust produced by the vehicle, and so can calculate the mass of each pollutant gas. Measurement of fuel used: Crucially, the drive cycle test gives a direct measure of the total amount of fuel used during the test cycle, and hence the fuel consumption. The annual inspection type test gives no indication of fuel used, other than the crude (and frequently wrong) assertion that "less ppm HC must mean better fuel consumption".
In general, these drive-cycle economy measurements show good consistency from test to test. However, in order to actually prove that a fuel "saving" device is of benefit, you need several tests both with and without the device. The next few paragraphs give an introduction to the statistical problems you need to be aware of, and is just as applicable to performance tests as economy tests. Variability: consider these two data sets:
A-B-A tests: again look at these data sets:
Some makers of fuel "saving" devices claim you can't use an A-B-A test because their device permanently changes the
engine (cleaning carbon deposits, or adding some kind of "magical" coating). In this case the best option would be to
have a "control" vehicle as well, identical in every respect except that it is fitted with a "dummy" device rather
than the device being tested (and ideally, the person doing the fitting should not know which is the real device and
which the fake). The two vehicles are then run through the same sequence of testing and road driving
as required - if the "magical" coating really works, the effect should be like this:
This sort of test program is of course relatively expensive, but a drop in the ocean compared to the vast profits that could be made from selling a genuine fuel saving device to car manufacturers. More than enough incentive for the makers of the device to carry out such rigorous testing, you would have thought.
Appendix: Variability of on-road and drive-cycle fuel economy data
The main reason for trusting drive-cycle data above on-road measurements is due to the very much better repeatability. For
example, in 1993 the Ecotek CB-26P was evaluated on the road by the Warren Spring laboratory. A
team of experienced scientists drove a vehicle repeatedly round the same test circuit, trying to maintain the same conditions
each time and so characterise the economy without the device fitted. The results (taken from the
report) were as follows:
(Runs 9 - 12 were with the device fitted, so are excluded). By contrast, take a look at a series of drive-cycle rolling-road tests carried out by the EPA on the Tail Pipe Cat:
It is immediately obvious that the variability of the drive-cycle tests is very much less than the variability of the on-road tests. The statistical measure of variation, "standard deviation divided by mean", is 12.6% for the on-road data and 2.1% for the drive-cycle data. In other words, the drive-cycle results are six times more precise than the on-road results - and remember these results were taken by a group of experienced scientists specifically trying to obtain good quality data. This gives a concrete illustration of why drive-cycle rolling-road data is the only sort of proof that industry will accept for a proposed fuel "saving" device.
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
