It's dark. You're driving down an interstate highway in the dead of winter and it's 20 miles to the nearest town. As you drive into the night, you ask yourself: What would I do if the fuel pump suddenly conked out? Or the water pump? Or the alternator? Could I walk that far in sub-zero temperatures? Do I trust my car that much?

For most automobile owners, the answer to that last question is evidently a resounding "yes." Judging by the number of individuals who drive their vehicles in Minnesota during winter, or Arizona during summer, or in any locale where help is hours away, consumers must believe in the sturdy reliability of today's vehicles.

True, some vehicles are undeniably more reliable than others. But as a group, the world's auto manufacturers are producing sound vehicles today. And that's true for the once-maligned American car, as well as European and Japanese models. "The reliability of U.S. cars has improved dramatically during the past ten years," notes Rana Arons, a spokesperson for Consumer Reports, which polled more than 604,000 owners in that magazine's annual reliability report.

In truth, the task of building a reliable automobile is one of the toughest in engineering. Unlike machine tools or home appliances, automobiles must run in a wide variety of environmental conditions, ranging from ice and snow to sand and dust. They face an extraordinary array of temperatures, from -70F in northern Canada to 130F in the Arizona desert. Owners drive them over gravel roads and washboard concrete. And unlike commercial airplanes, they aren't piloted by trained experts who know exactly what the vehicle can and can't do. Worse, many cars are poorly maintained by owners who seem oblivious to their requirements.

Given this wide range of variables, it's a wonder that automobiles are as dependable as they are. Part of the reason for their success is sheer effort. Ford Motor Co., which has made tremendous strides in quality and reliability in its recent Taurus/Sable models, has 25 full-time reliability engineers in powertrain alone. More than half of those have Ph.D.s. General Motors and Chrysler also have scores of engineers dedicated to reliability issues.

What have the automobile industry's experts learned about designing for reliability? How can those principles be applied to the design of other products? Following are the most commonly mentioned do's and don'ts related by reliability engineers and experts from university engineering programs.

Understand interaction failure modes

Mismatch two perfectly good components and you may create an unreliable system. If a starter motor lacks the necessary torque to quickly fire up an engine, for example, it will soon wear out. Why? Because longer cranking wears it out.

Similarly, heavy window glass may reduce cabin noise, but it can also wear out the power window motor regulator. "There are a lot of components that work fine alone, but will cause problems if no one considers how those components go together," says Steve Daleiden, a product design reliability implementation engineer for Ford Motor Co. (Dearborn, MI). "For that reason, your time is often better spent looking at how parts interact, rather than redesigning the same part over and over again."

Be aware of program timing

One of the biggest causes of poor vehicle reliability is missed schedules. Why? Because engineers who fall behind, often find that they don't have enough time to fix potential problems. "When you have a machine that consists of 3,000 components, it's absolutely critical to have overall awareness of your schedule," notes Bob Higgins, assistant brand manager for Buick's LeSabre brand team. Are the dies on time? Are the subsystems getting validated? Are the system assemblies passing internal tests? If you don't know the answer to these questions, Higgins says, reliability will suffer.

Understand what customers expect

Most reliability engineers say this is the key to designing any system. Twenty years ago, automakers could produce an engine that lasted eight years and customers would feel satisfied. Today, the same eight-year engine would be deemed a failure.

To better understand today's customers, Ford powertrain engineers attach flight data recorders to engines and record how severe-duty fleets use them. The recorders measure such variables as accelerator pedal movement, crankshaft revolutions, and velocity.

Similarly, the duty cycle and useful life objective for a four-wheel-drive vehicle will be far different than that of a luxury car. Knowing the differences between the duty cycles of various products is critical to creating a reliable system. "It's a matter of understanding what the customer wants and then being able to put their desires into engineering terms," explains David Patel, a reliability implementation engineer for Ford. "If you don't know what customers expect from the product, you can't meet their needs."

Use common systems and methods

Don't use one method for one department and a different method for another. In the automotive world, for example, engineers often design for a 95th percentile customer. But if they design for a 95th percentile customer in a suspension, then they must do the same for a powertrain and the interior. Such consistency holds true for all processes. Mathematical tools, such as reliability growth curves, for example, should be employed universally or not at all.

Communicate

More than ever, design engineers today need to communicate. To fully appreciate interaction failure modes, they must first communicate with engineers in cross-functional areas. Similarly, they must stay in contact with manufacturing engineers. Lack of communication can cause design and manufacturing problems, which ultimately ruin reliability. Many experts believe that lack of communications between design engineering and other disciplines was a contributor to the reliability problems in cars two decades ago.

Because many design engineers grew up in a "throw it over the wall" environment, however, broader forms of communication have been slower to catch on in some companies. "The old internal boundaries are vanishing," notes David Cole, director of the Office for the Study of Automotive Transportation at the University of Michigan. "To be successful, engineers need to accept the fact that they can't compartmentalize as much anymore."

Track vendor performance

"If a fuel pump goes out, as far as the customer is concerned, the automaker has failed them," Cole says. "They don't care if the problem can be traced back to the supplier." For that reason, manufacturers need to be concerned not only with the so-called 'Tier One" suppliers, but with suppliers' suppliers, as well. Automakers recommend that manufacturers keep lists detailing the performance of all suppliers, from Tier One on down.

Design for easy assembly

Systems that are difficult to assemble will more often be misassembled. By designing for easy assembly, engineers can improve product consistency, and vastly reduce the possibility of failure down the road.

The same holds true for easy serviceability, engineers say. If an engine needs to be hoisted out of the car to change a spark plug, for example, service is likely to cost more and customers are more likely to procrastinate. Result: reliability problems.

Consider external influences

Products too often fail because engineers haven't anticipated all the external factors that could influence their operation. Automotive engineers address these issues in the simplest possible way: They operate vehicles under the worst conditions. Ford engineers often test engines and ignition systems in Bemidji, MN, during winter. Most manufacturers also have proving grounds in Arizona, where their vehicles can operate under the worst possible summer conditions. "We try to draw the line at the 95th percentile duty cycle," Whitney says. "The engineering challenge is to determine where the 95th percentile duty cycle is."

Don't invent during product development

Unless a process is very disciplined, there's more risk in using an all-new concept. New products and components can be employed if their field performance is well understood. But once product development has begun, all processes and technologies should have already been "bulletproofed."

Regard reliability as a moving target

As long as a product remains on the market, engineers should monitor its reliability. If such efforts don't improve that particular product, they may help improve other products down the road.

Reliability experts often point to Japanese cars as an example of the changing nature of reliability. When Japanese cars first reached the American market, they say, those vehicles exhibited terrible problems with quality, fit and finish, reliability,and durability. Within a decade, Japanese car makers dominated all the reliability and quality polls. American car makers are now passing through a similar renaissance.

The key to ongoing success, experts say, is to constantly hone the development process while keeping an eye on reliability. "It doesn't matter if you design and develop a refrigerator or an automobile or any other product," Whitney concludes. "The same thought processes and principles apply to any kind of device."

Putting consumers first

Our readers say that the number one feature they require in a new car is reliability. So what aspects of it are automobile designers working to improve? Engineers at Ford cited the following four areas as most important:

Dos and don'ts of reliability

DO...

DON'T...