"I've calculated the variations as proposed in the article and they are less than the variations that we measure."
I guess this explains as well as anything else about how credible your conclusions are, as the article doesn't mention anything about dealing with variation due to sample size, samples being representative, etc.
The article just proposes how to deal with some variables affecting metal concentrations in oil, and chooses to ignore others. A metric is proposed, but it specifically mentions that 'statistically significant' samples are supposed to be used, and leaves the definition of significant to be addressed elsewhere.
In summary there is nothing in the article that you presented to address the problem of 'stastistical significance', but you seem to think that it does and as a result seem to have missed the point completely. The article shows that there is so much variation engine to engine that one cannot distinguish the groups of engines from each other, instead one would need to consider each engine on it's own, as is stated. This is similar to what has been suggested in the 'article of the month' posted for the site.
http://www.obitet.gazi.edu.tr/makale/makale/internalcombustionengines/047.pdf
4. Evaluation parameters
The compensated wear rate obtained represents the wear behaviour of an engine in a defined period of time, but the consequent question is “What level of wear rate is normal or abnormal for the engine studied?”. Answering this question will be the final target of a condition monitoring system for predictive maintenance based on oil analysis. At this point, it is important to take into account the important factors mentioned before: engine age, metallurgy, type of service, environmental conditions, etc.
A comparative parameter is proposed to answer this question. The parameter is defined as:
Z = EwrE¯ wr / EMwr, (14)
The reference wear rate represents the normal wear rate values for a determined sample population (from an engine or engine model). These are defined as statistically significant from statistical studies with a large population of samples. These reference values are not static and they evolve with the number of samples analysed and take into account certain restrictions [10].
The expression of Z takes into account not only the deviation from the engine reference wear rate, but also by using the engine model reference wear rate, compares the current situation to a larger population behaviour such as all the engines of the same model. The value is also transformed into a non-dimensional magnitude.
With sufficient historical data, limits for Z parameter can be set by the user (fleet owner, oil analysis laboratories, etc) to reflect maintenance goals. Hence, maximum benefit can be gained by using oil analysis as a predictive tool for condition monitoring.
I guess this explains as well as anything else about how credible your conclusions are, as the article doesn't mention anything about dealing with variation due to sample size, samples being representative, etc.
The article just proposes how to deal with some variables affecting metal concentrations in oil, and chooses to ignore others. A metric is proposed, but it specifically mentions that 'statistically significant' samples are supposed to be used, and leaves the definition of significant to be addressed elsewhere.
In summary there is nothing in the article that you presented to address the problem of 'stastistical significance', but you seem to think that it does and as a result seem to have missed the point completely. The article shows that there is so much variation engine to engine that one cannot distinguish the groups of engines from each other, instead one would need to consider each engine on it's own, as is stated. This is similar to what has been suggested in the 'article of the month' posted for the site.
http://www.obitet.gazi.edu.tr/makale/makale/internalcombustionengines/047.pdf
4. Evaluation parameters
The compensated wear rate obtained represents the wear behaviour of an engine in a defined period of time, but the consequent question is “What level of wear rate is normal or abnormal for the engine studied?”. Answering this question will be the final target of a condition monitoring system for predictive maintenance based on oil analysis. At this point, it is important to take into account the important factors mentioned before: engine age, metallurgy, type of service, environmental conditions, etc.
A comparative parameter is proposed to answer this question. The parameter is defined as:
Z = EwrE¯ wr / EMwr, (14)
The reference wear rate represents the normal wear rate values for a determined sample population (from an engine or engine model). These are defined as statistically significant from statistical studies with a large population of samples. These reference values are not static and they evolve with the number of samples analysed and take into account certain restrictions [10].
The expression of Z takes into account not only the deviation from the engine reference wear rate, but also by using the engine model reference wear rate, compares the current situation to a larger population behaviour such as all the engines of the same model. The value is also transformed into a non-dimensional magnitude.
With sufficient historical data, limits for Z parameter can be set by the user (fleet owner, oil analysis laboratories, etc) to reflect maintenance goals. Hence, maximum benefit can be gained by using oil analysis as a predictive tool for condition monitoring.