We recommend Internet Explorer set to 1024x768.

© 2003 Brian F. Schreurs
Even we have a disclaimer.

Just to keep you good and confused, we've collected a bunch of oil change interval theories for your consideration.

Theory 1: three months or 3,000 miles

Well, it seems to work. But for most engines it's a lot like changing your bath water partway through your bath. More than likely you're wasting a lot of time, money, and oil, but you're certainly not likely to be harming the engine any. A 5,000-mile interval is gaining popularity and is probably more realistic, but it's still a guess.

Theory 2: change it when oil analysis says so

This is the only theory not based on some degree of guesswork. An oil analysis will reveal exactly what kind of shape the oil is in and will tell you whether it's time to change or not. The problem is, an oil analysis costs as much as an oil change with conventional oil, so by the time you've paid for the analysis you might as well have just changed the oil. While oil analysis has other benefits -- such as identifying problem areas before they damage other parts of the engine -- it certainly is the most expensive of the theories here.

Theory 3: change it when the dashboard light turns on

Many cars now come with an oil change indicator light on the dashboard; GM has been particularly pushy about installing one in their cars. Vehicles with this indicator may no longer have traditional normal/extreme duty mileage recommendations for oil changes, stating instead to change the oil when the light comes on. But is it trustworthy?

Those indicators, at least as installed on GM vehicles, don't actually run tests on the engine oil. Instead, engineers program the computer with a lifespan for the oil under ideal circumstances. This ideal scenario varies from one engine to the next, as some engines are harder on oil than others. Then the computer "penalizes" the longevity of the oil based on service. GM believes the most critical external factors to oil life are its running time and its operating temperature; therefore, the indicator tracks engine rpm as a measure of running time and engine temperature as a measure of the oil's environment, applies mathematical penalties to the ideal scenario based on these readings, and blinks the light when the countdown hits zero.

While that's a fair bit of theoretical modeling, it's no worse than some of the other theories presented here, and also has the benefit of being tailored to the specific engine in the car. In our own study, the end result has been pretty good: during the Mobil 1 study the indicator activated at just over 5,000 miles, premature for synthetic oil but a reasonable expectation for conventional oil.

Theory 4: Paradise Garage Method

Here we attempt a very simplistic model to determine oil life based on the observed history of that oil in that car, based on the idea that TBN is the first critical measure of oil to fail. It uses the following equation:

In this equation, tested TBN is divided by virgin TBN to arrive at a percentage of remaining TBN. This percentage is then applied to the amount of tested miles to determine how many miles remain for the TBN. Then, the tested miles are added to the remaining miles to arrive at the oil change interval since the last oil change.

The advantages to this method are: it makes no assumptions as to the type of oil or driving conditions, relying entirely on observed test data; and it is simple to use. The disadvantage to this method is that it assumes TBN degradation is linear, which it is not. Therefore, the model is highly inaccurate at lower mileage intervals.

Try our method out for yourself:

This method, developed by Ted Kublin at Dixie Synthetics, uses the virgin oil TBN with a calculation of engine stress to determine the oil change interval. Here is the equation:

This equation divides cubic inches by horsepower as an indicator of the engine's natural stress level -- a high-output engine will be harder on oil than a low-output engine. Then it multiplies that against the miles per gallon of the specific engine being tested, as a representation of the environment that specific engine must work in (higher fuel economy represents less stress). Then it multiplies these by the oil capacity, as a greater oil capacity means the work is spread over a greater volume. Finally, the equation multiplies these by ten times the virgin oil's TBN (representing the oil's ability to handle the stress loads) to arrive at the oil change interval.

This method tends to emphasize long drain intervals, which with today's oils in today's cars are often quite reasonable. Nevertheless, it would be a good idea to approach extended drains with a certain amount of prudence. Kublin prefers to think of the TBN as an empirical constant and recommends using a TBN of 12 for Amsoil, 8 for Mobil 1, and 4 for conventional oil, regardless of its actual TBN value.

Try Kublin's method for yourself:

Paradise Garage reader Gary Heidebrecht presented us with this interesting, if complex, equation. It takes the available TBN in the engine oil and divides it against a mathematical model of TBN consumption to determine the oil change interval. Here is what it looks like:

Does your head hurt just looking at it? The beauty of this equation is that it lets you select an endpoint -- chances are, you don't want to change the oil when the TBN hits zero; it's better to leave some buffer in there. Blackstone Labs recommends a TBN of two. The shortfall of this equation is that you must include make-up oil in the equation, so you'll have to fiddle with it to compensate for your engine's oil consumption (for example, if the model says you can go 10,000 miles, and you know you'd add two quarts of oil in that many miles, then you need to re-run the calculations with an additional two quarts of oil to arrive at a more accurate number). Another disadvantage with this equation is that you'll need to know some engine specifications -- these are normally printed in the owner's manual or the service manual.

The numerator in the equation determines the amount of TBN available for the engine to use; it's straightforward enough. The denominator determines the rate of TBN consumption by multiplying a blowby model with a neutralization constant. The blowby model basically measures the "length" of the wall-to-ring gap and multiplies that against the applied force of compression, using the compression ratio to represent force. This blowby model is multiplied against a neutralization constant to determine the rate of TBN consumption.

The neutralization constant is key to the entire method. This is the rate at which TBN is consumed based on the amount of probable blowby. It is only fair to note that this is a constant in theory only -- in the real world, it's impossible to isolate every variable and arrive at a true constant. But, this number should be close enough for determining an oil change interval. If you've already conducted a long-term oil analysis, it's easy enough to determine the neutralization constant. Start by determining the rate that TBN units were consumed across the test:

Then determine the blowby model based on the engine specifications:

After that, the neutralization constant is the TBN consumption divided by the blowby model:

And, once you've determined the neutralization constant with a case study, you can solve the equation for TBN consumption without running an actual test, which is what happens in the denominator of the Heidebrecht Method.

Our Mobil 1 test resulted in a neutralization constant of 0.00000716. A friend's 1996 Volvo 850, which went 20,000 miles between oil changes, had a constant of 0.00000893. This difference, small though it may be, added 3,000 miles to the Camaro's oil change estimate when we used the Camaro's constant instead of the Volvo's. But enough with mind-numbing theory. Try it for yourself:

Conclusions So Far

Here's how the theories are stacking up against reality:

Theory five, the Kublin method, shows promise. Using the actual Mobil 1 TBN in the calculation presented a very reasonable number for an engine that consumes small amounts of oil, and the calculation using the empirical constant of 8 for Mobil 1 is almost spot-on for an engine that uses no oil at all. Nice work, Kublin!

Theory six, the Heidebrecht method, is also surprisingly accurate provided that you know approximately how much make-up oil you'll use over the course of your planned interval. Clearly, in the earlier stages when we didn't really know how much make-up oil we'd need, the estimate was way off. But by the end of the study it missed our actual interval by only 700 miles. So, for engines where the rate of consumption is known, it may be possible to get a reasonable estimate of how long the oil will last. Kudos to Heidebrecht for puzzling out this complex formula.