We recently provided examples of how we employ our text analytics and data visualisation methodologies to focus our efforts when seeking to establish whether the opportunity exists to reduce the costs associated with our condition monitoring programs.


In this case we looked at oil samples, however, the text-analysis and data visualisation process can be used in many other forms of data associated with maintenance programs (e.g. failure mode analysis and work orders).

As we alluded to in our last blog post, when analysing oil samples, we don’t just stop at the text. We also support our decisions with a rigorous review of the associated numerical data.


We find the most effective of these is a combination of Rate of Change (RoC) limits coupled with oil analyte (e.g. elements Fe, Cr etc) levels to complete a rigorous review of historical data for specific machine types and components.

As an example, review the following graph in Figure 1. It represents the change in iron (including PQ) and chrome over time in an excavator’s left rear final drive.


Keynotes supporting the plot are:

  1. Colored markers and bars represent whether the sample is an “A” (blue) or “B” (red) sample.
  2. Bars identify whether the oil was changed when the sample was taken.
  3. The dark shaded circle/ellipse (1 and 2) represent general notes. The clear circles (3, 4,5and 8) represent normal rates of change, yellow circles (7) represent low RoC while red circles(6) represent high RoC.


It must be noted that oil change is one of the most crucial considerations in this analysis. If the oil has been changed after the previous sample was taken, the real RoC will be higher than the slope indicated in the plot.


In this case, the combined RoC/analyte level analysis shows the following:

  1. Even though there was a low rate of change detected between circles 4 and 5, the level of iron, chrome, and PQ, OEM limits were exceeded (more than likely due to the 1000 hours between samples) and sample intervals were reduced to 250 hours.
  2. At circle 5 the oil was changed and while the plot shows a reduction in iron, chrome and PQ levels (at circle 6), the RoC is high because the oil was clean at the beginning of the sample period, which lasted only 250 hours.
  3. At circle 6 the oil was changed and while the plot shows another reduction in iron, chrome and PQ levels (at circle 7), the RoC is rated low because the oil was clean at the beginning of the sample period, which lasted only 250 hours.
  4. At circle 8 the RoC is normal and only PQ is above the OEM monitor level. A final conclusion as to any issues with this component has not yet been reached as additional information is being sought on any repairs/investigations. However, the above demonstrates that a combination of RoC analysis and analyte levels provides a significantly different perspective on oil analysis than analyte levels alone.

GETTING THE MOST FROM YOUR OIL SAMPLES - Relialytics

Figure 1 – Excavator Rear Left Final Drive Fe, Cr, PQ RoC and Limit Analysis


As a final note, the data at circle 2 is compromised due to issues with information provided on the sample card. Even though there are spikes in iron, chrome and PQ at this point, the severity is in question due to a number of unknowns.


This reinforces the importance of accurate data collection when taking a sample, the subject of an earlier Relialytics blog.