Historically, users of oil analysis have relied almost exclusively on commercial oil analysis laboratories or oil suppliers to identify when a machine is in alarm. Realizing the limitations of this approach, world-class organizations are taking charge of their own alarm settings to ensure that their specific objectives are met. The advent of sophisticated oil analysis software has put this objective within reach of most anyone who desires it.
The primary purpose for alarms or limits is to filter (funnel) data so that the technologist spends his time managing and correcting exceptional situations instead of laboriously perusing the data trying to find the exceptions. The alarm serves as a “trip-wire” to tell the analyst that a threshold has been passed and that action is required. Some data parameters have only upper limits such as particle counts or wear debris levels. A few data parameters employ lower limits like BN, additive elements, flash point, oxidation stability and FTIR (additive). Other data parameters like viscosity and FTIR use both upper and lower limits. These generally relate to important chemical and physical properties of the lubricant where stability of these properties is desired (not too much, not too little).
Alarming techniques vary to fulfill the requirements of different oil analysis objectives. These techniques can be generally categorized as follows:
Proactive alarms alert the user to abnormal conditions associated with controlling the root causes of machine wear, operating faults and lubricant degradation. They are keyed to the proactive maintenance philosophy of setting targets and stabilizing lubricant conditions within those targets. A strategic premise of proactive alarms is that they be set to levels that will generate improvement over past performance (e.g., cleaner, dryer, cooler, etc.) or ensure that conditions are maintained to within levels that have previously been optimized relative to organizational objectives. Within the proactive domain, we utilize the following types of alarms and limits.
Goal-based Targets – Targets apply to the control of parameters like contamination to achieve machine life extension (see Figure 1). For example, a hydraulic machine running at ISO 18/15/12 cleanliness will experience a three-times life extension if the fluid is cleaned to an ISO 15/12/9 cleanliness. Setting the limit at ISO 15/12/9 is a goal-based initiative. Conversely, if the same hydraulic machine is running at ISO 15/12/9 and control is lost, allowing the contamination level to reach ISO 18/15/12, we can expect a three-times increase in wear during that period. The desire to return the system to an ISO 15/12/9 is driven by a specific objective and is, therefore, a goal-based limit. This type of limit is usually applied to particle count, moisture level, glycol level, fuel dilution, AN and other common root-cause conditions.
Figure 1. Goal-based limits are used to reduce stress (contamination, for example) on the oil and machine to extend service life.
Aging Limits – Another type of proactive limit or alarm relates to the progressive aging of a lubricant or hydraulic fluid (see Figure 2). From the moment a fluid is placed in service, its chemical and physical properties transition away from the ideal (i.e., those of the new formulated oil). Some properties transition very slowly, while others transition very dynamically. Limits keyed to the symptoms of lubricant deterioration are referred to as aging limits. Aging limits can be effectively applied to such parameters as AN/BN, viscosity, RPVOT, LSV, PDSC, elemental spectroscopy for additives, and FTIR (oxidation, nitration, sulphation and additives) and dielectric properties. Figure 3 shows some example limits for both goal-based and aging parameters.
Figure 2. Aging limits alert users to the approaching end of the service life of the oil or machine component.
Figure 3. Example limits for both goal-based and aging parameters
Predictive alarms signal the presence of abnormal machine conditions or the onset of wear and failure. They are aligned with the goals of predictive maintenance – i.e., the early detection of machine failure symptoms as opposed to failure root causes (proactive maintenance). In oil analysis, a properly set predictive alarm has many advantages over other predictive maintenance technologies and, as such, represents an excellent complement to vibration analysis, thermography, etc. Within the predictive domain, we utilize the following oil analysis alarming techniques.
Rate-of-Change Alarms – Rate of change alarms are typically set to measure properties that are being progressively introduced into the oil, such as wear debris. The add rate (change) can be calculated per unit of time, hours, cycles, etc. For example, a 100 ppm increase in iron over a period of 100 operating hours could be stated as 1 ppm per hour of operation. When the parameter is plotted against time, the rate-of-change (add rate) equals the current slope of the curve (see Figure 4). Unlike level limits, rate-of-change limits ignore the absolute value of the data parameter, emphasizing instead the speed at which the level is changing. Rate-of-change limits are effectively applied to particle counting (unfiltered systems), elemental wear metals, ferrous density, AN and RPVOT. They also can be effectively applied to monitor abnormal degradation of additives with elemental, LSV and FTIR spectroscopy.
Figure 4. Trend-line slope is a visual indication of rate-of-change and parameter severity.
Statistical Alarms – For many years, statistical alarms have been used effectively in oil analysis. The practice requires the availability of a sufficient quantity of machine- and application-specific historical data from which to draw meaningful conclusions. The statistical alarming approach is simple. A population mean and associated standard deviation are generated from the available data. The data from a sample is compared to the mean of the population. If the value falls within one standard deviation of the mean, it is considered normal. If it falls outside of one standard deviation from the mean but within two standard deviations, it is considered a caution. If the result exceeds two standard deviations, the value is considered in critical alarm as it is higher, or lower as the case may be, than 95 percent of the population. Should the value exceed three standard deviations, it is a critical situation indeed, as the value exceeds the 99th percentile of the population.
Legend: U = Upper Limit, L = Lower Limit, P = Positive Slope, N = Negative Slope
Note: Where alarms are bi-polar, the first shown is the most important.
Figure 5. Table of limit types applied to data parameters
Figure 5 generally tabulates the applicability of different targeting and alarming techniques for specific oil analysis tests. It also identifies whether the target or alarm is upper, lower or both. Statistical alarming methods are commonly applied to ferrous density, elemental metals and other predictive oil analysis measurements. While very useful in oil analysis, statistical data can result in false positives and negatives due to poor stratification of the data with respect to machine type, application and operating environment.
This article was excerpted from the book Oil Analysis Basics - Second Edition.