Toyota recommends a particular interval for a timing belt replacement to avoid liability for failure and to avoid the adverse reputation that failures would generate. Here's how that scenario works:
Stage 1: Timing belts break and engines are destroyed. This happens on belts with less use than the recommended interval.
Stage 2: A meme develops. The synopsis is, "Don't buy a [fill in the blank]. The timing belt breaks and the engine blows up." No further thought is involved; memes are like that.
Stage 3: The meme expands. The synopsis is, "Your honor, my timing belt broke before it was supposed to be replaced according to Toyota. Their incompetence in recommending that interval cost me [fill in the blank], and I ain't the only one been hit by it."
Now, how would an engineer investigate the longevity of the timing belt to recommend the recommended replacement interval to avoid this? The idea is to make the interval short enough to prevent failures, but long enough to not deter sales. It's easy in principle, but difficult and expensive in practice. The steps are:
Step 1: Drive a group of test vehicles until the timing belts break. Note the mileage on the each vehicle when its belt breaks.
Step 2: Examine the distribution of mileages when failure occured. Very likely this would be a Gaussian distribution, which is characterized completely by its mean (i.e. its average) and its variance (i.e. its dispersion about the mean). The standard deviation is the square root of the variance.
Step 3: This is not intended to be a discourse on statistics, but the core of the method is quite simple. If the distribution really is Gaussian, then the probability of a timing belt not breaking if it is replaced at six standard deviations less than the mean time to failure is 99.99966%. The probability of failure is one in 294,000. So, the recommended replacement interval is set at least six standard deviations lower than the mean miles to failure.
Now, is this what Toyota did? Likely not. To use this method, one has to have enough data points to characterize the distribution of failures. That means testing lots of engines
to destruction in real-world use. That expensive. What's worse, it takes a very long time. As we've all seen here, vehicles have survived with original timing belts long after the recommended interval. Think about that carefully. The first generation Tundra could be out of production before sufficient sampling was complete, and yet we find a recommended timing belt replacement interval in the owner's manual of the first one off the assembly line back in 1999.
So, how did Toyota determine a recommended replacement interval? I don't claim to know. I suspect it was a combination of: 1) observing
wear vs. mileage on the belts of engines that did not fail; 2) comparing that to their own long history of producing engines with timing belts; and, 3) extrapolating a
mean time to failure and an associated
variance. That is
much less expensive, and it can be done very rapidly, as the mileage required is much less.
Now, what should you do, and why should you do it?
If you replace your timing belt at the recommended interval,
you should expect to find that it is not near breakage, because it should not be near breakage. If it was near breakage in the typical case at the recommended replacement interval, it would mean that Toyota vastly overestimated the proper interval. The fact that we don't see rampant breakages means Toyota didn't overestimate the interval.
The problem you face is that
you don't know if your own case is typical. The analogy to help understand this is that statistics can tell you, with great accuracy, how many people in your city will die of cancer next year, but it cannot tell you if
you will be one of them.
All you can do is gamble, and I don't gamble, so I use a simple risk/benefit analysis.
I expect to keep my '00 Tundra for about 250,000 miles, at which time it will be 24 years old. This requires
two timing belt replacements using the recommended interval. I have changed it twice. If I were to stretch the interval, I would derive
zero benefit unless I stretched it such that I changed it only once, thus saving the cost and labor of one change. Minimally, that would be an interval of 125,000 miles, and I would face selling it while it has a belt with 125,000 miles on it. If I still changed it twice, but at a longer interval than Toyota recommends, it would mean that whoever I sell it to would get more use out of that final belt and I would get less, which benefits him more and me less.
Bottom line: Lengthening the interval means
risk without benefit. I'm an engineeer, and to me, that is a poor tradeoff.
