Credit image: Pxfuel.com, Piqsels.com
Oil analysis has been used for several purposes, such as predictive analysis of failures, monitoring of wear and tear and determination of the correct interval for oil change. It is a valuable technique in several industrial sectors, since it allows to determine when an equipment should be intervened and what the detected failure is. But this is an area that has not changed much over time, although the techniques and instruments of analysis have advanced, the procedure as such remains the same; it is usual to take samples of lubricating oil from the equipment with a frequency determined by the hours of operation and send them to a chemical laboratory, an expert analyzes the samples and makes a judgment based on the results and his experience, then issues a report with which, whoever commissioned the analysis decides whether to make a stop for the maintenance of the equipment or continue with its normal operation.
Traditionally, laboratories and companies use manual analysis of oils, by which they require experts in the area to analyze the samples and issue recommendations based on the results. And because it is an analysis focused on the capabilities of the human being, there could be inconsistencies due to the difference of opinions among experts, without considering that the analyst who reviews hundreds of samples may experience fatigue and lack of concentration; which represents the possibility of not alerting in a timely manner a problem that leads to a failure in the equipment. In other cases, the equipment manufacturers provide acceptable levels of wear and loss of oil properties, and in the absence of knowledge of how the equipment will be treated, they use quite conservative levels to protect it, but in practice this causes an alert when no defect of the equipment or the lubricant is found yet, leading to the early shutdown of an equipment and the consequent expense of both labor and materials unnecessarily.
With the advance of AI these and other problems seem to have an innovative solution. Autonomous learning techniques use both laboratory data and equipment failure analysis data to recognize the true warning signals at the right time. This is because it is a technique that takes into account the complete data set over the lifetime of the equipment, instead of depending on acceptable ranges to issue a recommendation.
By analyzing several data at once, the AI can issue a very specific recommendation. By using autonomous learning algorithms and by means of time series analysis techniques to evaluate the evolution of machine components, the AI is able to warn about the failure of a bearing due to changes in the concentration of metallic elements in the oil or to recognize in a spectroscopic analysis the presence of silicon and to associate it with the ground input at motor through the air filter. Most of the data is collected in real time while the machines are in operation, thus minimizing the impact on the production process.
Equipment maintenance professionals know that the oil in a machine wears differently according to the working conditions; the AI asks questions and learns from this experience, accumulating an immense database of conditions and parameters based on the historical results of the laboratories, taking into consideration any factors that contribute to the warning of a potential problem.
To conclude, we are witnessing how Artificial Intelligence increasingly impacts on the maintenance of industrial equipment, showing in the post an area of predictive maintenance in which the AI, using an autonomous learning based on data generated from oil analysis, combines these data with those from sensors on the equipment and compares them with the historical failure to detect potential failures in the equipment monitored in real time, this allows to take timely corrective actions in the equipment that require it.
I believe that, with the increase in telematics in equipment and online monitoring we will continue to see a greater volume of data generated, which represents an opportunity to combine this data set with those generated in analytical laboratories and thus improve predictive analysis techniques.
With the arrival of artificial intelligence in industrial maintenance activities and with autonomous devices, a world of possibilities is opening up that will make many companies jump on the AI train.
Thank you for reading me. Until the next post!
This post has been published previously in my other blog.