Executive Summary
TLT readers are upbeat about the potential of technologies such as online sensors, Big Data, artificial intelligence and the Internet of Things (IoT) to positively impact oil analysis in the next five to 10 years. But those same readers caution that while increased speed of report and reduced long-term costs are alluring, these technologies need significant improvement before gaining widespread acceptance. With online sensors, higher accuracy remains an elusive goal, and readers are reluctant to commit to them fully. Said one: “I am still an avid proponent of laboratory analysis as the ultimate yes or no for a fluid.” Additional sticking points for these systems include reliability of the data and finding qualified data-analysis personnel to interpret the results. However, while many unanswered questions surround these technologies, TLT readers remain enthusiastic about their potential role in OA. “It is the future and cannot be denied,” stated one.
Q.1. How will online sensors and offline laboratories impact oil analysis in the next five to 10 years?
I think laboratory analysis is going to be needed to confirm and diagnose the cause of abnormalities found by sensors, but sensors will become the primary screening tool.
Both have their place, and this is unlikely to shift strongly anytime soon. The bigger business will be contracting the interpretation of data from either source.
While sensors are important, relating the output of said sensor to the oil condition is a critical need. What constitutes a change in oil condition—Debris? Dieletric and acidity? Viscosity? If so, what does that mean for the performance of the oil—film formation, heat-removal capacity, shear degradation behavior and any other functions of the oil in the system?
Huge increase as it allows end-users to get real-time data that correlates to the oil analysis program.
Massive! A game changer.
I think we are going to observe an increase of online sensor usage, more so on applications or devices with high operational and maintenance costs. On the other hand, I believe offline lab capacities will be so impressive they will be a sure complement for online sensors on those expensive devices and a regular policy for all the other classical systems.
Online sensors could be used as an indicator to control quality but could fall short if there is a dispute or claim. Bring in the offline lab analysis to validate the data produced by online sensors.
Invaluable.
The marriage of the technologies will depend on reliably correlating them. If not correlated, then sensor data will likely remain independent of offline laboratory data similar to thermography and vibration data.
While online sensors provide instant analysis, they do not replace lab analysis. They have limitations relating to which tests can be run, and they do not necessarily follow standard test protocol like ASTM or ISO. Also, in most cases this kind of data is not provided to the outside lab to be considered in the historic data analysis trending.
For online sensors there is still a long way to go with respect to condition monitoring. However, such activities are necessary.
Online sensor may play a deep role for pre-detection. Oil analysis may be rediscovered as for fine analysis and understanding of the phenomenon.
They will become an ever-increasing area of development and implementation as more fleets and large users, especially in remote areas and for less-experienced personnel, look to cut costs (and corners) in their lubrication budgets by extending drain intervals to the last possible moment. Component costs, also increasing due to global supply strains (higher wages, material and energy costs and government taxation), will increase significantly, and maintenance costs will rise.
Online sensors are very good in most applications from the articles we have reviewed. In most new installations, there is room for the capital investments to provide an ROI. In most equipment that has been in service, in some of our cases 60-plus years, there is little room for the capital investment when historical OA programs have been very successful in preventing failures and long-term wear. In our case with very little cost due to services provided by our lubricant vendors.
The trend in the industry is more and more automation. I expect online sensing will increase. However, whenever an online sensor detects say a water contamination above the max. allowable, operations will request a manual sample and confirmation analysis.
Modest. Technology is limited to, for example, viscosity or dielectric values. Oils are more complicated than the analyses can provide; you need multiple oils analyses to really get to the detail of the oil performance degradation. And interpretation. Might be that through machine learning we can make a step forward, but I don’t see it happening readily.
Online analyzers will continue to grow in use. Offline laboratory analysis will continue but become more of a confirmatory check of the online analysis.
I think we will see some new technologies, like varnish detection in hydraulic tanks and systems, starting to appear. Also, with IoT, more of remote monitoring or automated systems might become useful and viable. As long as sample handling and result evaluation remain dependent on skilled operators and lab personnel, this will be but a gradual change.
There will be an increase in online and machine-side analysis. However, there will probably still be a lot of systems where it is more economical to sample and analyze offline in a lab.
Online sensors for industrial equipment such as rotary air compressors still fall short of being able to measure viscosity, acid number and water for a reasonable initial cost of sensor equipment. Laboratory analysis could grow in popularity as synthetic fluids continue to warrantee the life so long as periodic testing is used.
Online sensors have improved over the years and give great trend analysis of oil systems. I believe they will continue to improve and offer better trending data in the years to come. However, I am still an avid proponent of laboratory analysis as the ultimate yes or no for a fluid. Laboratory data gives the most accurate picture of the status of the fluid and is your ultimate double check for any online sensor.
I think in-line oil analysis will continue to digitize and become easier to use at a cheap cost with a broader mainstream acceptance of smart phone- and application-based monitoring.
With the rapid advance in online sensor technologies and fast mobile network, more oil analysis will likely be done online and in real time. That might affect the offline lab analysis business.
Online sensors will play an important role in some applications but are unlikely to be the only line of defense in critical operating equipment. For critical equipment, online sensors will be an important verification tool for lab analysis.
Online sensors will eventually become more accurate and therefore more widespread, but it will be beyond five to 10 years from now. For the next five to 10 years, offline lab analysis will continue to be the preferred method of testing.
I think this technology will become more common with some customers and likely more common with OEM-installed systems.
I am hoping more will move to online sensors.
Ideally online sensors would allow remote monitoring, but the reliability of those sensors/measurements will be critical. Robustness of the equipment and process will be key. Until this happens, offline laboratory analysis and data reporting will be used even though it will not allow for speeding up timeliness of data reporting.
Online sensors might spot an immediate issue; however, the offline lab offers analysis that a sensor may not provide. Offline labs will still be utilized for a long time due to low cost, availability and analysis of data.
Online sensors will start to be implemented as a quick and reliable tool for oil diagnosis and supporting decision on oil drain. Offline laboratory analysis will see increasing oil sample intervals, so less oil samples to be tested.
Their usage will increase. The tipping point for when we see widespread adoption of online sensors will be based on the price point per sensor. Once large-scale production results in a price drop greater than 10 times, we will see strong growth in their use.
My impression (basis on my experience with computers) is the correlation will not be very good.
I believe online sensors could most likely play a leading role; however, I believe having a backup system of analysis should provide greater QC assurances.
For oil sensors to provide usable information, they need to fine tune their capability to the level of a lab instrument. If sensors actually provide usable information, there is still a need for interpretation, correlation and trending. If the machine’s electronics are unable to provide this, then labs have a great opportunity to offer that service. Additionally, labs can confirm the sensor findings.
I feel that the IoT will bring a lot of new and useful information to the online game and allow for real-time monitoring of critical assets. (I am trialing a couple of different options this month.)
More online sensors in stationary machinery, but rolling stock will still use outside labs. Offline labs will still play a part to verify readings of online sensors.
As the IoT grows, so will this technology. I believe it will be more of a screening tool to warn of possible problems and not have the sensor capability of a full laboratory.
Temperature has a long history of being tracked online. There are many other avenues where sensor technology is exploring applications. Optical sensing may prove the most cost effective and also able to track changes real time.
Online analysis is coming step by step. Sensor technology is still the challenge. Particle analyzer with water percentage and temperature is one fairly well-working one, but also it has certain limitations as we know. Offline laboratory analysis is anyway needed for references and calibrations.
Online sensors will replace offline laboratory by 10%, but surely online sensors will dominate the market.
As technology improves and costs come down, online will be used more.
A large amount of time and money have been invested in advancing oil analysis into the real-time realm. Great strides have been made to develop sensors that would enable complete real-time monitoring of lubricant condition. Although there are many challenges remaining before an all-encompassing real-time oil analysis device exists. OEM recommended trade/brand itself need to be further tested to determine their true operating capabilities. Finally, research needs to be conducted to establish the feasibility of combining online and offline analysis systems to provide an all-encompassing real-time oil analysis module.
With technology advancements, online and offline analysis will become more prevalent. Quicker results, especially at the point of sample, gives the customer an understanding of the health of the equipment and the health of the lubricant.
I think both will get more emphasis and use, with significant developments in the accuracy and value of the sensors.
Just the human error factors involved in self-collection demand a significant change in how we monitor our in-service lubricants. This will require direct, real-time connection to the lab with results being immediate and automatically produced.
Online sensors will become more sophisticated/accurate. Offline lab analysis as a check for the online date will continue for older equipment.
Within five to 10 years I think oil analysis will be done more by online sensor because it saves time compared to offline analysis thereby helping to avert big problems.
Yes, good to have this online sensor. Lower cost is the big issue.
More and less. More online sensors, less lab analysis.
Oil analysis using online sensors and offline lab analysis is going take place but not in the next five to 10 years. It’s already started and is quite beneficial in estimating life of product.
Online sensors will improve, and their usage will spread rapidly. Offline analysis will recede gradually but not dramatically.
If sensors take advantage of low electronics prices and start doing simple and cheap applications, they can start being a useful tool (especially considering remote applications emerging in renewables or automotive electrification). Development still is needed as relatively little evolution has been done in past years compared to evolution of electronics. For laboratories, with the development of base stocks and bioproducts, I am not expecting many changes. More work still needs to be done to adapt standards and reclassify products to better and more easily fit industrial needs.
Online sensors for fixed plant will and must become part of AI in the future as they will contribute greatly in the reliability and productivity cost of any plant. Having real-time data available on your in-use oil will support decision making in maintenance intervals and preserve a costly resource to its maximum live. Today we waste a lot of usable products by interval-directed changes. For mobile equipment there needs to be a lot of work done still to make online sampling a reliable option. Mining is an example.
I think online sensors still have a huge path to cover in order to be a real alternative to offline laboratory analysis. But considering a five to 10-year span, maybe things will evolve and the online sensors will improve.
Online sensors should increase usage as mobile technology evolves, and I believe this is a very good screening tool to detect problems and trigger more in-depth lab analysis.
The pressure for labs to reduce their price points has resulted in poor data consistency. Use of online data will expand.
I don’t think online sensors will do what old-style lab testing could do in providing a clearer picture of the analysis.
In the next two years, will you be installing more, fewer or the same number of online oil debris monitoring sensors?
More
44%
The same
50%
Fewer
6%
Based on responses sent to 15,000 TLT readers.
Q.2. What do you think is the potential for Big Data analytics and other emerging technologies to be applied to oil and debris analysis?
Historically, we have a lot of data on our equipment but no tools to analyze it as a whole. With emerging software tools to help agglomerate data and provide insights, we will potentially be able to better target and predict impending issues.
Yes, but costs will be high.
There could be significant advantages in industries with mobile fleets (transportation, construction, mining) that tend to have fixed drain intervals coupled with common sump sizes but not so much with industrial plants (fixed assets).
Big Data analytics offer valuable modeling for any system. But we must remember that the results are only as good as the data provided. As long as we are consistent in our sampling and results, these tools can greatly help us see wear in systems and prevent catastrophic failures.
I think at some point there is too much information that just gets in the way. We go from looking at particles counts to ferrography and isolating these particles, which is great to learn what it is and where it might be from, but at some point we still need people in the field listening to and understanding their equipment from outside a laptop.
There are too many variables.
The potential would depend on the value offered and customers’ willingness to adopt these new technologies.
This is increasingly likely to be a useful tool.
Improved diagnoses due to larger data populations.
Where I see the most need for used oil analysis is with tests like analytical ferrography. This is a powerful tool that is underutilized in most instances; it can be very revealing when part of the overall test regime.
As soon as online sensors are readily available, there will be a huge benefit together with Big Data analytics.
As long as there is available WiFi coverage in all areas for remote monitoring, then more companies will see the benefits of reduced downtime from trend analysis of their machine’s condition.
If they can produce a user-friendly format that does not overburden end-user workloads, I think there will be many successes—if they can keep costs to reasonable levels. Reporting software that is efficient in detailing issues in a manner that would report at an executive-level viewpoint would be very beneficial. Typically, it is a 30-second glance to see where conditions are not acceptable, then on to the next issue. We have internal reports that were created to do so but were very time consuming to develop. We engineering folks love the stream of data; for senior leadership, ‘Be good, be brief, be gone,’ is the scenario across many industries.
It may be useful with certain monitored oil properties. But including data in a large data lake that may be several years’ worth of data may result in unreliable trending information.
The same oil will degrade in different directions if you apply the same conditions. Call it chaos theory, call it chance. On a macroscopic level you can see the oil degrade, but it might result in higher oxygen content, viscosity growth, aldehyde and ketone formation, acidification or any combination of them. Add to that metal contamination, chemical degradation—it’s complex and complicated. I think technology can deal with complicated, but how do you teach a machine to deal with complex? I remain unconvinced.
Big Data analysis will grow into a tool of choice to identify data trends with used oil analysis data across multiple machines and locations.
There is significant potential. The main gains will be for highly critical equipment (where the ability to pick up sudden changes is important) and resource-intensive but low criticality monitoring (where reactive maintenance is most efficient).
More data equals better analysis equals longer lubricant life and better payoff for implementing an OA program.
Definitely. Big Data and AI will play increased role in oil and wear analysis to provide insights that otherwise would not be available.
OEMs have been using Big Data for decades to establish maintenance recommendations and requirements, as has ASTM for setting test standards. Big Data is great for applications like this, but not so great for identifying outliers. Big Data analytics combined with human expertise will continue to be very powerful.
Very limited. The return versus the investment is not there.
Seems to be a focus for most vendors as we need something new to try to set ourselves apart from competitors. Technology seems to be the future.
It will take some time but automation and trending live will assist with flagging contamination and wear earlier.
There’s good potential for Big Data in terms of defining patterns to predict oil quality degradation, anticipate oil life and evaluate engine wear and engine set up running parameters.
Not enough samples and also not enough knowledgeable people who submit samples for analysis and know how to interpret. Big Data needs to be applied with consideration to the level of knowledge and correctness of tests requested and samples submitted.
The potential is there; getting people to buy in to the cost is the hard part.
I work for an oil company and our lab is already testing this technology after two years of development. The system will be heuristic and improve with every sample imputed and diagnosed. This will be a game changer for oil analysis.
Yes. However, it will benefit companies who gathered data for a longer period of time for data analytics or data-driven decision.
Most of these are more smoke and mirrors rather than providing real value.
A complete oil analysis system when combined with data fusion and automated reasoning could provide a huge boost forward for lubrication system conditioned-based maintenance. Oil analysis should therefore be a priority to any maintenance in order to ensure that all machinery is able to run as smoothly as possible. Further advanced oil analysis equipment solutions also are incredibly valuable for ensuring the continued health of heavy machinery. Not only will this improve machinery efficiency, but it also serves to reduce downtime. This can have a dramatic effect on output and costs, especially when combined with wear debris analysis and other condition monitoring methods. From a larger perspective, reliability and maintenance optimization are increasingly using information on lubricant condition in their methods concerning maintenance. Oil analysis has proved to be an effective tool for determining failure modes for both equipment and the lubricant.
The lab rats will find crude oil evaluations that will improve oil manufacturers margins.
This is an opportunity to develop a new industry-specific profit center for some ambitious young engineers.
Big Data analytics will be introduced to lubricants, but it won’t gain too much recognition within the next 10 years. It will take much longer to accumulate enough data to correlate sensor recordings with the equipment functions and failures. Failures do not happen very often, and they will be analyzed by technical personnel, not AI at least for the next 20 years.
It is quite limited. From my experience, rules and capabilities obtained are exactly or really similar to the ones well-trained and experienced personnel have been using for years (and you still need personnel to take the right decisions).
With Big Data analytics comes the challenge of inter-connectivity between the various data bases of the world users to really become a workable table of application and learning. The basis of input data into the BD process will require a guaranteed baseline quality to ensure dependability and confidence in the source.
The potential is huge. However, the path to having reliable models based on Big Data analytics may be quite difficult for such applications. I’m skeptical about Big Data models that aren’t based on the physical phenomena of machines.
I think that, like other areas of technical development, more and more improvement will be made. What the customers do with this data is the question.