Reliability Demands a Plan

Aug. 1, 2007
This is a condensed version of reliability consultant Ricky Smith’s article posted at’s sister site Written from the end user viewpoint, you’ll find the full text at

By Ricky Smith, contributing editor at our sister magazine Plant Services.

Management policies often tell Purchasing to buy at the lowest cost. Engineering is told to meet a deadline no matter what. Vendors are ordered to deliver equipment on time, at the lowest cost, with no requirements for a specific level of reliability. Maintenance is urged to get the production line up and running quickly.

These policies are designed to save money and time. But as they take effect, companies find asset reliability is becoming less controllable, or they’re spending large amounts of money to preserve it.

Most equipment failures fall into the infant-mortality category. In the world of asset reliability, we define “infant mortality” as the failure of an asset during startup, within a short period of time after installation, or soon after the equipment has been overhauled to a previous-state condition. This short time between when equipment is started up and when it fails could be months, or minutes. Reliability studies typically show about 68% of known failure modes are a result of infant mortality.

As most equipment is operated, it becomes less likely to fail. At some point, the probability of failure levels out to a random failure plateau, which allows failures to be detected through a proactive maintenance strategy. Infant mortality is difficult to identify and detect before failure occurs.

The U.S. Department of Defense (DoD) completed research on how to make equipment more reliable, so that once it’s commissioned or overhauled, it will have a high probability of operating failure-free for a specific time. The standards are at

To design and purchase new equipment with the highest probability of meeting your reliability needs, you should focus on a few simple questions.

First, what’s the overall expected downtime of the new production line or process? Years ago, I developed a lifecycle planning process for all new production equipment and production lines for a large international firm. The easy part was determining how much downtime could be allowed because we knew the production rates. The hard part was determining how much of the maintenance downtime would be unscheduled vs. scheduled. We established that 90% would be scheduled.

Second, what is the required mean time between failure (MTBF) for new equipment to meet the business goals of the new process or line? If companies would research and purchase programmable logic controllers, motors, gear reducers, and even specialized equipment with known failure rates, the production lines would have a much higher probability of meeting the reliability and production goals set by their corporation.

Organizations in the U.S., Canada and Europe have determined reliability rates and set standards. The primary U.S. organization is the American National Standards Institute (ANSI). If a manufacturer states its equipment conforms to ANSI specifications, you can be assured the reliability data is, in fact, reliable. If an equipment manufacturer doesn’t have MTBF data on its products, consider whether you’re willing to take a risk on reliability by purchasing it.

During the early ’90s, I was asked to visit a facility where a large gear-reducer failure on a critical asset had shut the plant down. How long had this gearbox been in operation? “More than 20 years.” Had the gearbox previously failed? “No.” The gearbox had simply worn out.

A local vendor recommended a “better” gearbox they had on the shelf. I told the company I would purchase the same type of gearbox that had failed and stay away from the “better” gearbox.

Having the right data at the right time could help a company save millions of dollars. One must set policy that directs purchasing to provide MTBF data to engineering or maintenance before a purchasing decision is made on a specific type of equipment.

Ricky Smith is a reliability consultant and a contributing editor at Plant Services. You can reach him at [email protected].