Machinery Lubrication magazine published by Noria Corporation
Issue link: https://www.e-digitaleditions.com/i/359954
www.machinerylubrication.com | July - August 2014 27 included. The statistical approach should deliver reasonable limit values. The viscosity distribution is bell-shaped. The median and mean are in the same range. SPC and the standard deviation can be used to identify alarm values. Limit values already exist for this parameter based on the ISO viscosity grades. In this case and for this special oil, the limits could be defined more precisely according to the statistical evaluation. The evaluation of the additive molybdenum also shows a bell shape, but the distribution is slightly high-reference skewed and bimodal. The median and mean are different, which indi- cates that the cumulative approach should be chosen. The population should also be investigated to identify samples with special cause variation. Figure 7 contains warning values for iron in different wind turbine models. It illustrates how statistical methods can be much more precise. These statistical-based limits can be combined with trend-based limits to help determine whether an increase in iron is within the acceptable range when compared to the previous sample. Limit Values for Hydraulic Final Drives Another example comes from hydrostatic final drives. These systems are used as crawler drives for excavators and agricultural machinery. Oil samples are often taken from these systems during the oil drain procedure with static intervals between 1,000 and 2,000 hours. The main purpose of the oil samples is to confirm that the right lubricant is in use, contamination levels are below permis- sible limits and wear rates are normal. In general, the evaluation of oil analysis results for this application is suitable for abso- lute limit values. Figure 8 shows the distribution of iron for a crawler drive. Results for different models were from the same manufacturer, and for all samples, a comparable oil type was used. The main goal of the statistical evaluation was more precise wear limits. Figure 9 shows the results of the cumulative method. Unfortunately, the new statistically based warning limits differed significantly from the actual limits in use based on experience. The new calculated limits seemed too high, and the unusual wide range of iron distribution supported this estimation. A closer look at the last 20 percent of the distribution revealed that the data population contained a large number of varia- tions associated with uncommon causes. Two independent but major effects were identified: a loss of viscosity due to contamination with hydraulic oil and high silicon content, which indi- cated high dust levels due to damaged seals. Figure 10 shows the statistically based limits for newer models from the same manufacturer. The wear limits are significantly lower and corre- spond with the previously experienced wear levels for normal conditions. The two problems of the older final drive generation have been solved for the newer generation. In conclusion, it is important to keep in mind that oil can talk and that limit values are an essential tool for evaluating analysis results from used oils in order to rate a machine's wear, oil condi- tion and level of potentially harmful contaminants. While limit information can be provided by a variety of sources, in some cases it may not be available. General or global limit sets also are not usually valuable for condition monitoring. Depending on the application, absolute or trend-based limits, or a combination of both, often deliver the best results. TYPE # SAMPLE METHOD FE-GREEN FE-YELLOW FE-RED A 610 1 (CUM) <835 835-1850 >1850 B 774 1 (CUM) <1508 1508-2313 >2313 C 1528 1 (CUM) <864 864-1331 >1331 TYPE # SAMPLE METHOD FE-GREEN FE-YELLOW FE-RED D 2128 2 (CUM) <75 75-160 >160 E 1291 2 (CUM) <98 98-264 >264 F 565 2 (CUM) <364 364-475 >475 FiGURE 9. Limits influenced by special cause variation FiGURE 10. Typical iron limits for final drives