Executive Summary
TLT asked these same questions in the February 2014 Sounding Board and wanted to see how the answers compare 10 years later (available at www.stle.org/files/TLTArchives/2014/02_February/Sounding_Board.aspx). While some readers anticipate little change in oil analysis, others expect automation and artificial intelligence to make a big impact, but also introduce new challenges.
Q.1. How will oil analysis change five years from now?
Online sensors will be the biggest change.
To be combined with online sensors and artificial intelligence (AI).
The existing tests will be dramatically updated to leverage new digital capabilities.
Onsite for rough analysis, automation for typical lab work.
Become more important to understand the status of in-use oil, particularly with lower viscosities and changes in the chemical box.
Oil analysis usually provides wear debris analysis, metal content and oil properties change to determine the life of the lubricant. Increasing automation and AI will make it faster and easy to display the results and predict remaining oil life.
Maybe AI will be used to process the vast amount of used oil analysis data.
Awareness of importance and impact.
Improving use of AI in analysis. A better understanding of the binary divide between sick and healthy machines.
I believe that AI will be integrated more with decision making, which will be more consistent than human interpretation; however, my belief is that it may be better 90% of the time; 10% of the time only a human can make proper judgements.
Not sure it will other than possibly more ferrography provided as inductively coupled plasma (ICP) and other data is already there.
Truthfully, I’ve been in this business nearly 30 years now and I haven’t seen any real significant changes in that time. So, based on that trend, I’d have to offer I don’t see any real significant changes in the short term.
Probably more real-time answers on equipment and oil condition with inline instruments reading and sending data through WiFi.
The oil analysis online could reach more reliabilities five years from now.
AI will be incorporated in analyzing and interrupting the data.
I believe more places will do their own analysis on site.
Online oil monitoring would be nice, but I don’t think that is going to happen. It will probably stay the same.
More automated methods; shorter turnaround times.
Oil analysis will be much more widely used as customers understand the significance and the cost savings aspects available.
Oil analysis, just like everything else, will continue to become more exact and detailed. Smaller, more affordable analysis machines that do more tests at a faster, more accurate rate may hurt the third-party analysis business.
Engine oil will become 100% synthetic versus natural crude.
1.) Using more onboard sensors and analyzers. 2.) Abandon the old bottle sampling method.
Progress is slow and depends on the availability of new methods and equipment. No significant changes are expected in the next five years.
In my opinion, online devices will provide a complete picture of oil and equipment condition. Most probably infrared (IR) spectrum measurements and models derived from AI will let us a precise equipment condition monitoring. But this alternative will be a feasible solution just for high cost or critical assets.
Very little, if at all.
Online monitoring will be most seen.
More and more online oil analysis tools will be used.
Oil analysis will be integrated into advanced pattern recognition models and AI analysis as part of a software package that is synergized with other predictive technologies.
Obviously the venture into the future will include more exotic synthetics with very different typicals. I tend to think more silicon-based fluids to manage heat will be expanding. I also believe mineral-based synthetics will be implemented in the future so used oil analysis will need to keep up with the diversity of available Group III and Group III+ base oils. 100% Group IV PAOs are dropping off due to performance of Group III base oil.
I don’t think oil analysis will change much five years from now. Five years is too short of a time to see the evolution of changes.
Folks will attempt to apply AI. This is likely to be a disaster in many cases, as needed information (and likely past experience to draw from) is typically lacking. AI’s tendency to “make up” things it doesn’t know will still need to be confirmed via a reliable source.
Oil analysis data will be important and will be shared with multiple users such as site technicians, management and engineering to decide on the equipment’s health. Employees will access other sister stations’ equipment health for comparison purposes. This will lead to a reduced number of lubricant types used, and this will reduce the logistics that will benefit stations economically.
As the testing industry is getting commoditized, we might see less of manual labor and more of automation and robotization.
AI will become a much larger factor in how and when you actually do your oil analysis. Collecting and analyzing system information through AI will be a key tool to monitoring your oil and grease health.
Probably more and more automatic testing devices; however it is important to have the experts and experienced people for the right diagnosis.
Perfect place for AI to analyze and predict performance.
More inline, real-time analysis.
More online sensors will be in use to monitor current conditions and alert to potential problems.
More on site for immediate results.
I do not see big changes. Basically the only change is the availability of more specific analysis.
There will be more sensors and readily available data. Not all data will come from traditional labs, nor will there be very long turnaround times for samples.
It hasn’t significantly changed in the last five years (dare I say 10) other than computerization. I guess AI programming may make its way into analysis of results and the drawing of conclusions.
The trend will continue to go toward using condition monitoring to predict failures before they happen. As oil analysis evolves, users will be able to minimize or eliminate unnecessary downtime by replacing oil or replacing/servicing parts before they reach failure. Oil analysis will allow users to become preventative rather than reactionary.
Increment in grease testing.
Far more analysis will be undertaken on site with immediate results.
I believe additional tests will be developed, especially around varnish formation and elimination, and current tests will be further developed and become more accurate.
The accelerated path of AI will help improve the commenting algorithms now in place for most laboratories.
AI will likely be a factor, if it’s not already.
More accurate and faster results.
The amount of different analysis methods will increase.
Online oil sensors and Internet of Things (IoT) devices will be more prominent, but traditional oil analysis techniques like ferrography can never be replaced with AI.
Probably through the emergence of real-time, connected onboarded sensors.
There will be more monitoring with sensors at the site.
Bit-by-bit automation and remote monitoring is increasing within sensor development.
More online measurements. Integration of the results in a data management system to correlate oil data with other process data.
Inline monitoring of oil and greases will take off. Therefore, the oil will be continuously analyzed.
Oil analysis is likely to undergo several changes over the next five years, driven by advancements in technology, industry trends and environmental concerns.
The capability of inline sensors will improve to be able to carry out more in-depth analysis in real time.
Increasing automation of collection and diagnosis.
We will see increased utilization of sensor technology, which will provide real-time information on lubricant parameters and wear.
I think that AI is going to help a lot, especially in the analysis of information and recommendations to be delivered to the end-user.
Tribology of oil will remain the same, but there may be many new frontiers in oil analysis at delivery ends.
It will simply advance to further automation. Electrification may add parameters to the oil analysis.
Develop the preventative maintenance, extending lifetime of machines and extending lubricants utilization.
I think new methods will come up to assess varnish and efficiency.
Sensor readings will start to be incorporated into the analysis process.
I hope more will use it!
Q.2. What is the most important issue in analyzing oil analysis reports?
Understanding what the limits are and where the metals in the report come from in the equipment.
The report is not always related to the operating conditions or lube oil manuals of the equipment.
An easy to understand, actionable summary.
Quality of the data and the interpretation.
If you are consistent with one lab, you are certain of the information you are reviewing. If there are multiple labs generating the reports, then there is always the question of reproducibility and repeatability.
The health of the machine that the oil is lubricating. Oil chemical change and wear debris analysis provide evidence.
Understanding the working conditions of the machine being analyzed.
For most understanding what they are looking at. Also depends on oil types being used and the reports as things are different for different oils.
Two-fold: 1.) Having a technical aptitude with regard to the equipment being sampled and 2.) offering beneficial and useful recommendations to the client without talking over their heads.
Have accurate information on the machine and oil type.
Reducing failure time.
Finding trends and data outside of limits.
Sorting out what is important for the system being analyzed.
Sampling. Take a sample that is representative to oil in service, handle it with care and store it in a dry and clean bottle.
Determining if the data is relevant to the application and what the data’s integrity is.
Not using one data point or one aspect of the report. It is important to look at all aspects and allowing the analysis to paint a more complete picture.
Trending the concentration of additives and wear metals along with the ISO numbers.
Understanding your own situation and taking that into consideration when evaluating a report.
Engine oil temperature at which carbonization of molecular structure of the oil occurs.
1.) Understanding the results and how to implement solution. 2.) Key guide for understanding the data.
Correct interpretation of the laboratory results.
Knowledge and experience.
Trends, actions to be taken, wear rates.
Important to arrive at a decision by looking at various parameters.
Recognizing your critical parameters for each component.
The biggest issue is that despite decades of oil analysis experts reciting the “garbage in, garbage out” mantra to our clients, they continue to send in samples with incomplete, inaccurate or entirely missing information. In order for our data analysts to provide an accurate report, they absolutely must know the type of component along with the make, model, oil brand, type and grade, hours on the oil and the component, etc. With online sample submission, it is easier than ever for clients to enter of all that information online one time, and then rely on the lab’s systems to retain and retrieve the pertinent information, but they still can’t or won’t make the effort in many cases.
The purpose of analyzing oil analysis reports is to determine the health of the application. The most important issue, then, is to determine what relates to the health of the application. It then becomes straightforward to determine the health from the history.
The weak technical education of the customer.
Viscosity, particles (ISO 4406), their aspect and their nature and elemental analysis.
Having a baseline (new oil) analysis for comparison. Historical trending analyses also are important. Historical information like application, machine hours, maintenance history and oil fill/change history also can be missing, but generally can be backfilled, while historical samples cannot be obtained once sampling opportunity has been missed.
The most important issue is the time delay between the oil sampling date and the release or approval date of the oil analysis certificate or logistics problem because of the distance between the plant and the laboratory. The second challenge is when the result of the analysis arrives the engineer has to make a final decision, enter the information and attach analysis results to multiple sites for tracking the trend.
Inadequacy in expertise to assess patterns that can imply underlying problems.
Understanding the source of contaminates found in your used oil report in order to address the underlying issue.
In the trending of the data, it is imperative that we understand when a “bad” sample appears. That data can mislead people and allow poor decisions to be made.
Oil condition as well as systems trend based on correct monitoring interval, application category specific.
Proper data interpretation.
People often do not read “routine” analysis reports. I think those are the reports you want to read closely—the equipment can still be saved. With respect to a reportable or critical report, that equipment is likely already damaged, and it often is already back in service before the results are returned and the unit fails in the field before you can respond. I think more time should be spent on routine sample results before it becomes reportable or critical. Watch for changes in trends versus absolute flags.
Paying attention to particle counts (are they up or down), presence of silt and degradation of antiwear additives.
To capture the contexts and correctly identifying the salient features and assessing the priorities.
Identifying and following trends when analyzing the same oil over time. Individual results may not tell a story/help understanding.
Getting the complete information from the customer that relates to unit make/model, compartment make/model, oil brand/type, time on the oil/unit and if there are any concerns or recent repairs. The more information that is given, the better able the analyst is to provide recommendations.
Correct data interpretation based upon the application and component.
Knowledge.
Change in viscosity, increased wear and dirt debris and change in acid number.
Best practice is conducted. Conditions are well documented.
Actually there are two most important items: 1.) knowing the characteristics of the fresh oil so that change from new to used is understood, and 2.) having baseline data trends for same/similar equipment that the oil sample was taken from.
Identification of limits violations and/or trends toward those.
Results interpretation.
Trending data accurately.
Having proper make and model information, time on oil and unit, oil type and viscosity, having previous sample data.
The condition of the oil and its remaining useful life (RUL) and level of contamination.
Understanding the data and interconnecting parts.
Ensuring that the sample is extracted correctly and properly labeled for unit type, lube type and grade. Without this information, the data analyst is handcuffed, and the recommendation loses some of its accuracy and value.
Complete information on the component (make and model) and component hours of service and oil service hours.
Accurate and consistent input at the equipment end.
Standardization across global standards, specifications and requirements enabling conclusive quantitative results.
Water and fuel contaminant ratio of the oil.
Accuracy of the tests’ data.
Change in metals content.
Test method and results.
Being properly trained in how to read and understand the analysis report.
Trusting results that do not seem correct.
To see actual condition and be aware about quality trend.
Electronically readable.
When interpreting the results, the actual use case of the oil must be taken into account. Extensive, detailed measurements mean nothing when they are not put into the right perspective.
That the correct limit values are applied for diagnosis and application-related recommendations are given. Clear structure of the different topics—oil condition, contamination, etc., and highlighting abnormal values.
Look at the trends and evolutions rather than focusing on non-systematic deviations, which may be due to errors (sampling or experimental).
The relation with the performance of the oil.
Real cleanliness of oil and assessment with microscope.
Only the standard measures. No look at spectra changes—only the regular peaks.
Accuracy of the data and rigorous quality control.
The most important issue in analyzing oil analysis reports is ensuring the accuracy and reliability of the data.
Understanding the use case. Ensuring test equipment is calibrated and verified for the results obtained.
Lack of previous results to evaluate tendencies.
Knowing what the challenges are the equipment is confronting in its operation for understanding what to look for.
Interpretation of the results and understanding correlation between things.
1.) Looking at trends. 2.) Understanding the potential shortcomings of the collection point/methods used.
For Latin America, the main problem is that the end-user dedicates enough time to analyzing trends in the equipment.
I look first at wear metals.
Truly understanding the value and insight that they are providing.
Do you think oil analysis is on the cusp of major changes compared to how it’s traditionally been conducted?
Yes
56%
No
44%
Based on an informal poll sent to 15,000 TLT readers.
Interpretation of data.
Comments and guidance.
To look at the overall picture and not to focus on single outliers.
Identify action to take to solve a problem in case of deviation, abnormal result.
Viscosity changes from initial viscosity and how much the acid number increases.
Having a clear goal for analysis. Is it primarily for uptime, diagnostics of failing equipment, extending drain life? What defines success for the program?
Many important properties, and they are all there for a reason. The most important issue is being able to spot a problem that the report is pointing out.
Q.3. What are the most common mistakes people make with oil analysis?
Not using it to set oil drain intervals.
Not following through on recommendations for remediation or even not reading the report at all.
Not understanding why they are asking for certain tests and not understanding how to use the data.
Not understanding the minimums and maximums for each viscosity and knowing when contamination is coming from internal components versus external sources.
Focus on the oil property rather than the health of the machine.
Cross contamination at the testing lab.
Not providing enough information.
Taking a bad oil sample, and not taking the sample under the correct operating conditions.
I can only speak about our program. and the largest mistakes made for us is accuracy of information and complete notes given to the lab.
Not having a reference sample to compare against and not taking good samples and confirming things. Need to react to reports where needed, not just file away.
Two big mistakes are not adhering to the first two of five critical success factors. Critical success factor (CSF) mistake 1: Pulling poor samples, many times allowing external contaminate into the sample. CSF mistake 2: Providing less than adequate sample information with the sample.
Incomplete information and then people expecting pinpoint accuracy in data and analysis.
Not considered the results trend and omitting the baseline for each component.
Not taking action.
Not identifying critical analysis for their systems and not notifying the proper people (communication) when a serious issue is reported.
Interpret the data without comparing to your baseline. Trends are more important than absolute numbers.
Improper sampling/non-representative samples submitted.
Holding onto the samples for a time before sending them in for analysis. Also, sample taking procedures are so important to ensure a good, representative sample that is not pre-contaminated or an invalid sample point.
Taking samples with a slightly different method every time will skew your results. Not reading or understanding the whole picture represented in the report. A lot of valuable data is left on the table that your company paid for.
Taking things too literally. Just because one parameter is “off” doesn’t mean that the oil needs to be changed out.
Engine oil requirements are not based on fuel (diesel, gas, natural gas, propane or hydrogen).
Submitting samples that are not representative of the oil in service, not using oil analysis as a trending tool, not following up on abnormal laboratory results.
Simply change the oil without digging into, identifying with evidence, then resolving root causes.
Interpretation and taking actions without liking for tendencies or just looking when alerts come out.
Picking representative sample and calibration of equipment by strictly adhering to test method.
Do not look at trends. Do not know how to read a report, and worst, most of them do not know the context of the system where the sample was taken.
Not looking at the results in depth. Too many folks look at the critical level: 1, 2, 3 or 4. If it is a 1 or 2, they just file it. There can be critical indicators in a 1 or 2 report.
Two mistakes are seen. The first is not performing the correct test, i.e., running tests that do not have an impact on the performance of the oil in the application. The second is not having a history of the analyses. A lot of times, there isn’t even an analysis of the fresh oil as a baseline.
To make conclusions based only in oil analysis before making a good RCA.
Trusting/acting on the results of a single oil analysis.
Appropriate sampling procedure not being followed, incorrect reference oil details, failure to monitor trends and historical data, incorrect limits.
Not looking at them in a timely manner.
People don’t match with other tests.
Reacting too quickly to bad data.
Sampling techniques or missing procedure to follow as well as correct data registration, e.g., oil age/top up and system operating time.
Relying on insufficient data.
Reading the comments portion of the report first. You should interpret the analysis report yourself and see if your conclusion agrees with the lab. The comments section of the report is the last thing you should read first.
Many labs just give a simple overall condition report (good, bad, normal, etc.) People tend to just look at that instead of reading the report and looking at trends and/or anomalies.
Poor sampling procedures.
Incorrect/incomplete data entry from the customer; assuming too much or too little from the data; the sample is a “check the box” effort in some cases, and the data is filed but not actioned.
They do not read up on the methods! Applicability, precision and bias. It is all there!
Not following through with recommendations from a report.
Taking the sample improperly and not providing enough information about the sample.
The way and consistency of how the samples are taken.
Not embracing the fundamentals of complete lubricant and equipment registration, sampling supplies, sampling procedures, program oversight and work to define resulting benefit.
Sample prep.
Ignoring the results of an analysis by not changing oil and/or continuing to run worn and/or defective machinery.
Wrong sample handling. Incomplete data about machinery condition. Those lead to wrong analysis report.
Taking first oil sample only when a failure has been suspected to occur. Thinking oil analysis will tell them exactly what failed and why.
Contamination, mislabeling, delayed analysis, measurement equipment calibration drift.
No trending, rely only on one analysis to take action.
Not providing enough sample information. Taking action on samples with abnormal or severe data.
Contaminated sampling tools and incorrect sampling techniques. Insufficient oil and component information, e.g., oil type, oil age, machine information, hours on oil, etc.
I believe most mistakes with oil analysis begin with proper sample collection. Garbage in, garbage out.
Not knowing what their goals are.
Lack of consistency.
Inability to properly interpret results.
Not aware of the repeatability and reproducibility issue.
Not comparing from last time.
Not following the recommendation for corrective actions we provide to our customers.
Taking a representative sample as well as providing the correct information describing the fluid, fluid age and machine details are basics that are all too frequently not performed well when sampling. Samples are sometimes not sent for weeks and perhaps months to the lab. Samples are taken too infrequently to build up usable trend analysis, and lastly the lab detects an issue, perhaps a serious one and reports this, but no action is taken by maintenance personnel.
Misunderstanding the data that is contained in the reports.
Not interpreting trends correctly, to maximize oil drains.
Generally they neglect external conditions and equipment health.
Overcomplicating things and not having an understanding of the actual application the analyzed oil is used in.
Starting from wrong sampling, using non-suitable containers, providing insufficient information about assets and oils.
Use it retroactively (only after a problem occurred) instead of proactively. Wrong sampling practice (frequency, method, location).
Only looking at oxidation changes.
Insufficient look at the samples themselves—some abnormal visual aspects are sometimes better alerts than measured parameters.
Incorrect sampling, oil contamination.
Not providing the correct sample. Looking at generic contamination limits to interpret specific cases.
People expect an accurate report about machine condition making only one oil analysis.
Cross contamination in the testing equipment.
Most common mistake, checking viscosity with thumb and forefinger and for particulate with oil in palm of hand rubbed with forefinger.
Assigning alert levels.
Sampling and reports ignorance.
People tend to forget about the need to evaluate trends and instead they focus on single analyses.
Decided by oil appearance to change the oil, not fully tested the key parameters. Sometimes the machined components surface finish and tolerances cannot attain and decided to change the oil.
Not providing correct/complete information about their samples.
Poor sampling, not providing clear and accurate information.
Not interpreting the results correctly, not maintaining an adequate history on the fluid or not documenting what the condition of the equipment was.