Many manufacturing companies have a lot of time and money invested in rotating equipment. They recognize that to protect their investment and achieve efficient, cost-effective operations they must keep these machines operating at maximum capability, which is accomplished by performing regular machine maintenance. Most machine maintenance falls under one of three maintenance philosophies: run to failure (RTF), preventive (PM) or predictive (PdM).
The most expensive method of performing maintenance is RTF, as it usually results in additional, and sometimes catastrophic, machine damage. Allowing the machine to operate until the bearings collapse could damage the shaft, wear the bearing housing, rub the impellers or worse. Additionally, the parts that usually fail aren’t typically stored in spares inventory. They carry a premium price and involve extremely high shipping and transportation costs just to get the necessary parts on-site. Add to that the cost of labor required to return these damaged machines to operating condition and it's easy to see how operating a maintenance program as RTF is the most costly of the maintenance philosophies.
The preventive maintenance philosophy is based on the premise that most failures are age-related and, by conducting minimal maintenance tasks at predetermined time intervals can prevent many catastrophic failures. Detailed studies have been conducted to determine the mean time between failure (MTBF), with maintenance tasks and intervals developed accordingly. In the mid-1980s, a reliability study determined that only 11% of failures were age-related, whereas 89% were random. In fact, 68% of failures occur either immediately after installation or after returning from maintenance, indicating that the failure probability actually increases by performing maintenance tasks under the preventive method. Also, the probability of failures increases with the complexity of the machine.
Figure 1. Easy-to-use tools, because they identify a problem, its cause, and a correction, have put electrical reliability at the forefront in most plant reliability programs.
The predictive maintenance (PdM) practice takes non-destructive measurements during a machine's operation to determine the machine’s condition. Use of this practice received a jolt with the introduction of the data collector in the mid-1980s. Predictive maintenance technologies perform standard “non-specific or specialized” measurements, and are generally non-invasive and easy to use. Predictive maintenance consists of three phases: detection, analysis and correction.
The detection phase identifies a problem machine by screening as many machines as rapidly as possible. Once the detection phase identifies a bad machine, the analysis phase is implemented. This phase might require additional measurements, tests or technologies, and is similar to getting a check-up by a doctor. The general practitioner does the initial screening, and if additional tests are required, specialists are called in. Once the problem is analyzed properly, the correction phase is implemented.
Many PdM programs fail because they omit the important analysis phase, or they attempt to take too much data during the detection phase. Going from detection phase to correction phase, leaving out the analysis phase, is analogous to having open-heart surgery because of high blood pressure. Furthermore, if the corrective action is based solely on the data taken to identify a bad machine (detection data), often the wrong corrective action is taken, wasting resources and losing credibility for the program.
Conversely, if too much data is taken during the detection process, valuable time is wasted unnecessarily. This limits the number of machines that can be monitored during each route and requires additional labor to survey the entire plant completely. This means some machines don’t get surveyed, and in a typical mature PdM program, it’s estimated that 2-3% of new problems are detected during each data collection route. Often, post-mortem forensics of a failed machine reveal that sufficient data had been taken to identify the problem, but the data had not been reviewed and, consequently, the developing fault was missed.
Procedures for the correction phase can range from continued monitoring at reduced intervals to a complete disassembly and rebuild. Therefore, the key to an effective, less costly, correction phase is accurate, thorough procedures during the analysis phase.
Today there might be several generating stations sending power to the grid supplying your facility. The power might travel through many different transmission paths and across several municipalities or distribution districts before reaching the plant. There are currently 50 different power quality standards in the United States, one for each state, and yet power supplies aren’t restricted to state boundaries. The bottom line is that the quality of the power used to drive a plant's multiple machine motors can vary.