CarTalk.com Blogs Car Info Our Show Deals Mechanics Files Vehicle Donation

How is the lifetime of a part determined

While looking through my owners manual it shows that certain things should be replaced at X miles. Like the timing belt, it says they should be replaced every 60,000 miles. So my question is how do they come up with that amount? I have been on here discusing other car parts and such and have been told something can fail within weeks of instalation depending on the part and such. So how do they come up with these time lines for replacement?

Reliability engineers perform “accelerated life” tests. Their test protocols subject the parts to all of the extremes and all of the cycles they’re required to meet in a highly accelerated manner. “Mean Time Between Failure” determinations are made. Parts that fail are subjected to analysis to determine their mode of failure. Many parts are pushed beyond their specification limits to determine how much abuse they’ll take and how they’ll fail when they ultimately do fail.

All of this data feeds both the design package and the support documentation, which includes the scheduled maintainence recommendations.

Another extremely critical area of analysis is called “Failure Mode Efects Analysis”, or FMEA. What it boils down to is if a parts fails, how does it manifest itself? What effect does it have on everything associated with it?

For example, if a spark plug wire brackets fail is may have no effect other than to reduce the life of the wire due to oscillations. Therefore, the testing it goes through might be minimal. If a piston connecting rod fails, it’ll destroy the engine. Therefore, it becomes critical to ensure that the design is robust and will stand up for many many miles to whatever stresses it’ll be subjected to. Ensuring that all rods MEET the design requirements becomes a whole 'nother field of Quality Engineering.

The subject of high performance parts comes up here occasionally, and that interests me. What people should understand is that high performance parts may not be as robust as OEM parts. Rather than long term reliability, their design goal is often light weight or enhanced power. Often this is at the expense of long term reliability or life. Slotted and drilled brake rotors, for example, may dissipate heat better under extreme use, but may eat brake pads rapidly. Race vehicles only need the pads for one race, not for 60,000 miles.

So basically it’s a guestimate? They say according to our studies we feel it’s best to replace part X at this point because the likelihood of failure beyond that point is good?

But shouldn’t there be like a range? I mean if they say 60,000 miles, wouldn’t it be better to say between 60 and say 65? Also is there a place we can go to see what the failure rate for part X is? Like does it fail 20% of the time at 30,000 and 55% at 50,000 and so on?

So basically it’s a guestimate?

 It is a little more than that.  They have far more experience with the car than you and I normally would in a life time.  They also will test parts that have been in service and use that to estimate the expected life.  

By the time they get to writing the owner's manual, they have a good idea of what to expect.  They will likely be rather conservative when it comes to timing belt life, as a timing belt failure can cost you a new engine. They may be less conservative with say the cabin filter.

Timing belts are a special case, because they are so critical to engine operation.

The manufacturer errs on the side of caution. The timing belt may last longer than 60K miles, but you’re gambling if you try to stretch it, especially if your vehicle has an interference engine.

Some manufacturers recommend longer timing belt replacement intervals, such as 70K, 100K, or more. It’s up to the engineers.

The important thing to remember, especially with timing belts, is the TIME interval, which is just as important as the mileage interval. Most owner’s manuals recommend replacing the timing belt at XX,000 miles OR XX months, whichever comes first.

WHICHEVER COMES FIRST. That’s the important thing to remember with timing belts.

Other replacement parts, like starters, alternators, etc, have no specified lifetime. They may fail in a week, or they may last many years.

My personal experience is that replacement parts last a long time. I have never gotten a bad replacement alternator, starter, or any other major part. Maybe I’m just lucky, but I’ve been replacing these parts, and similar, for more than 30 years.

It isn’t a guesstimate. It’s based on extensive testing.

And if it’s critical it usually includes a design “error budget”. A part in use might have to withstand a continuous load of 500 pounds with with a shock of 50,000 pounds for 10 milliseconds at a given ramp speed, all at -20F. If it’s a critical safety part, it migh actually have been speced to withstand 1-1/2 times that amount, and have been tested to exceed twice that amount before failing.

It’s a science. It isn’t a guesstimate.

A range would make no sense. The number, 60,000 miles, has been determined to be that point beyond which probably of failure begins to rise unacceptably.

I surely think there is alot of technical and mathmatical and materials calculations involved but I also think that the accumulated base of automotive know how plays a part. I guess what I am saying is we got to where we are out of lots of trial and error combined with science.

Things still suprise the engineers though as fuel pumps are showing up to be the weak link in direct injection. I would have thought all that could be known about a fuel pump is known but that was a incorrect conclusion.

Considering the amount of fuel pressure required for a direct injection system to work properly, the fuel pump design is probably in virgin territory. IIRC, the direct injection system requires fuel pressures at 300 psi and up. Much higher than the 5 psi for carburetors and 60 psi for the fuel injection system in my 2000 Ford.

A friend of mine’s parents purchased a BMW with direct injection. The system was plagued with problems, and after 3 fuel pumps and other fuel related issues, BMW bought back the car.

Some parts are a “guesstimate” and other parts are based on actual test results. Spark plugs can be tested since they operate in a very controlled enviornment. If the replacement interval is 30K miles you can expect close to optimal performance for those 30K miles, after that plug performance can drop off.

Timing belts might be more of a guesstimate since the life a belt can vary greatly from belt to belt. The belts are tested and if the motor works properly mechanically then the belt will be 99% percent likely to last the duration of the replacement interval. After you go past the replacement interval the percents start to shift. Say if you go past the replacement interval by 2 years your belt has an 80% survival rate. At 4 years perhaps 50% survive. So past the replacement interval you start gambling.

The only timing belt failure I ever had was at 27,000 miles on the first car I had with a timing belt. The car just stopped running at 20 mph. The only thing I knew was that I wasn’t getting any spark. It wasn’t until I had someone crank the car when I had the dist. cap off that I realized what had happened.
However I had a 92 Plymouth Voyager with a 3.0 v-6 that was supposed to have the belt changed at 60.000 miles and the job on that car looked very intimidating so I called the local dealer to see how often they really fail and was told they had never seen an in-use failure, all the belts they replaced were done because the water pump was leaking.
That car went to the junkyard because of rust after 14 years and 170,000 miles with the original timing belt and water pump.

My experience is as soon as it fails, it’s lifetime is over.

In addition to what has been said about testing, there are also techniques for monitoring the failures in the field. If an engineer has data about early failures, he can contruct a bunch of statistics predicting - fairly accurately - when the rest will fail.

So if you have a part like a timing belt, if you have reports of early failures, there are statistical techniques that can be applied to see if the problem is in the part itself, or just a statistical artifact. This is paryicularly effective in monitoring the effect a particular change might have had.

They do testing in as close to a real-world simulation as possible without putting the part on a car and running it for 15 years. For a shock absorber or suspension component for example, calculations are done on how many cycles (travels up and down) the part will likely experience, then an automated machine equipped with sensors pushes and pulls and bends the shock in every axis that it’s likely to experience force, and at forces similar to what it’s likely to experience, for however many millions of cycles they want the life expectancy to be. Typically they will actually test the part to destruction and analyze the failure mode, as mountainbike said. Further tests will be done to determine the maximum force the shock can stand, to analyze whether the seals stay intact and excessive corrosion occurs in heat, cold, mud, salt water, etc. And the part will be tested on an actual vehicle, for “NVH” or “noise, vibration, and harshness” and durability as well, albeit in a more limited fashion, or it would take many years for it to actually reach production. Sometimes a string of failures will occur, and a manufacturer will ship damaged parts back to the OEM supplier for failure analysis and further testing/improvement.

This is just for a mechanical part. Electrical parts will undergo torture and real-world simulation tests of their own, and a complex system like an engine, transmission, traction control, or an entire vehicle will need much more extensive testing. The problem is that no one can ever find every possible way a complicated system like a vehicle will fail in every case, and in the real world, someone will always do something unusual to the vehicle in a way that the engineers never thought of. Then there’s cascade failures, where any one factor wouldn’t cause a problem, but the combination of several small issues, sometimes in a critical sequence, will cause a problem that no one could have predicted years down the road.

Sometimes when a part fails, it may be the supplier’s fault and not the automaker. Sometimes it may be the automaker taking a supplier’s part and using it in an unapproved application that it was never meant for, or some critical detail was omitted by the automaker when the specs were given to the supplier. Or possibly the part was damaged during assembly. There are so many ways things can go wrong, and this is assuming there’s no intentional deception going on.

So I guess that the whole thing could be summed up as: “Engineers do the best they can”, “The world is an imperfect place; screws fall out all the time”, “$h!# happens!”, and “Make something idiot proof and along will come bigger idiots”

indy; you have been given some very knowledgeable and good answers. Manufacturers have a quality philosophy wich determines how long something will last. As mentioned, exhaustive testing will determine how long a part with a certain quality level will last. In the past the leaders in quality parts and durability were the Americans and Germans. Of the British only Rolls Royce had any degree of quality.

Then (in 1947) the Japanese realized that the only way to sell cars, watches, cameras, etc , in the West was with quality and durability. The two are not the same. Volkwagen and Audi have good quality (fit and finish) but crappy durability. Toyota, Honda, Mazda and Mitsubishi all jumped on the quality (low initial failure) and durability bandwagen and selected the best MATERIALS and the right safety margins to get maximum life. They also instituted 6-Sigma quality at the producttion line; this allowed about one bad part in 340,000.

At this time GM and others were allowing 1% defects, which the warranty process was supposed to correct. The result was amazing quality at a low price for Japanese vehicles and was the start of the downhill slide of US manufacturers.

Fast forward to today; of the Japnese manufacturers, nearly all have the same approach as Toyota, but even that results in some products (mostly electronic) slipping through the net.

Most US car makers now are applying Total Quality concepts to their design and manufacturing with promising results. Prior to that they were practicing “Value Engineering”, another term for cheapening things up, without reducing the life.

Deawoo, the Korean car maker obviously did few of these things, and their cars have both poor reliability and short component life. GM is now injecting some of the those concepts into the company and hopefully the new Aveos and other cars from that firm will be better.

In the meantime, Hyundai has taken heed of what it takes to design and build good cars and the sister company, Kia, is not far behind.

To sum up. if you want long life and low failure rate:

  1. Select a good design which is robust; the motors in German Braun appliances are a good example.

  2. Use long life materials, not plastic intake manifold gaskets, like GM

  3. Apply a brutal testing procedure to ensure the design is right

  4. Practice 6 Sigma quality control with both suppliers and the assembly process.

  5. Design a user-friendly vehicle that is easy to service, and provide realistic service intervals.

Of the above, few companies in developing countries do any of this. In the developed world, Fiat is just becoming familiar with some of these concepts. Russian car manufacturers typically only practiced #2 to some extent. US manufacturers are only now putting all these concepts in place. Too bad two had to go bankrupt to learn that lesson.

On a local but smaller scale, Hamilton-Beach, the small appliance manufacturer recently bought Proctor Silex of the “burnt toast” fame. The new company gives all products a 5 year warranty. This is a direct result of increasing the quality and durability of the products.

They want to be in the same league as Kitchen Aid, Cuisinart, and Braun, upscale appliance makers.