Reliability Engineering Snapshot TM

Illustrated Case Studies in the Maintenance Reliability Engineering World of Failure Analysis, Predictive Maintenance, and Non Destructive Evaluation



Lubrication - Case Study No. 32: Evaluation and Findings of In-house Oil Analysis

Wondering about buying some oil analysis equipment and analyzing the oil yourself? Doing oil analysis in-house, instead of sending the samples out to an independent oil analysis company, has its advantages and disadvantages. This is one company's long story.

Company "A" has one plant mechanic that does all of the lubrication. There is a printed schedule that the Lubrication mechanic, or Lubricator, uses to keep track of what equipment needs to be lubricated, and how often. The total number of lubrication chores that this Lubricator has to do in one year is 23,900 tasks. That averages out to one task every four minutes. Needless to say, the preventive maintenance aspect is just a bit out of control. It does not take a genius to realize that the Lubricator has taken it upon himself to make sense out of the ridiculous schedule and come up with his own. The only way to survive and stay on top of contamination is to do only that work which is warranted, and this is done by monitoring the condition of the equipment's oil using an in-house oil analyzer. The oil is changed only when the oil analysis determines that the oil is bad, and not just because it is "on the preventive maintenance schedule."


The oil sample is placed over an electric grid which sends alternating currents through the sample. The manner in which the alternating current changes is indicative of what kind of contaminant is in the oil, and how much of it there is in the sample being tested. Although the little oil analyzer is not in the same league as an oil laboratory, in this case, the information desired is simply knowing that a contaminant has entered the oil. One of the basic functions of the oil analyzer is to look for relative changes in the amount and type of contaminants in the oil. It can discern between ferrous and non-ferrous contaminants, and whether there is water present. It cannot differentiate between the actual metallic elements, therefore, a non-ferrous contaminant can be anything from sand to tin, or even nickel. This is the analyzer's weakness, when compared to the capabilities of an oil analysis laboratory. However, a sample showing a high non-ferrous content can always be placed under a microscope, and from there one can differentiate between sand and something else. The analyzer does a good job of trending the changes in the level and type of contaminants (i.e. ferrous, or non-ferrous).


In the early part of April 1997 the plant maintenance engineer and the Lubricator went to an oil analysis class for a week in order to learn about the new oil analysis software and hardware which was purchased by the plant. The plant oil analysis program began in earnest on April 25th, and was in effect for eight months. During this time 186 oil samples were drawn from 100 pieces of equipment within the plant, an average of six samples were taken each week. Since the program's inception, the oil on 54 out of 100 machines analyzed had been changed; and on half of those machines requiring an oil change, it sometimes took two to three oil changes to get all of the contamination out of the equipment. These same 100 machines had PMs requiring an oil change of only once every four years!

Making time to learn how to use the oil analyzer required that the Lubricator devote one full day per week. Another half day was spent obtaining about eight oil samples. During the first five months the maintenance engineer also spent one full eight hour day each week helping train the Lubricator on the computer, supervising the oil analysis, and debugging the software and hardware. The computer used for the oil analysis program was located in the office of the maintenance engineer; this facilitated training the Lubricator and answering his questions immediately as they were generated. During the last three months, the maintenance engineer reduced direct supervision of the Lubricator down to one hour each week because the Lubricator was well versed in using the oil analysis software and became self-directed. During that hour, the maintenance engineer would review the ongoing corrective actions with the Lubricator, and the results of the current oil tests. Both parties would then agree upon a course of action for the following week. This type of interaction was very effective in addressing oil problems and prioritizing oil changes.At the end of three years, the Lubricator was averaging 12 hours to complete the acquisition and analysis of 21 samples each week, self-directed.


The primary failure mode of all of the oil had been contamination by water, and the secondary failure mode had been contamination by the process product. Not one oil sample had deteriorated from natural causes due to the extended age of the oil and its subsequent depletion of additives. The root cause of the contamination was the seal design and oil breathers. Ordinary labyrinth seals and lip seals are designed to keep oil in, and are not designed to keep water, steam, and other contaminants out. Even "bearing isolators" cannot withstand a pressurized stream of water coming from blown pump packing, a driving rainstorm, or even from a well intentioned washing down with a pressurized stream of hot water. A stream of steam constantly impinging upon the bottom of tank agitator gearboxes will fill them up with water very quickly. And those oil breathers, they are more trouble than they are worth; the home made ones with just a plain open ended elbow pointing downwards, are the worst. One might as well put a funnel on top of the machine. Those style breathers were systematically eliminated.


MYTH No. 1 - "You know you have water in the oil when it looks milky, if it looks clear then its OK."

Not so. By the time water can be seen in oil each water droplet has grown to approximately 30 microns in size, that's because the human eye can't see anything smaller than that size, and therefore the oil looks clear. Water droplets as small as one micron have been found in oil samples, and on average their size is five microns. The oil analyzer can detect this size water droplet and this has been confirmed by visual inspection under a microscope. If the droplets don't coalesce, and continue to accumulate at the smaller size, then the oil's additives can be used up and the oil may in fact turn corrosive, all the time looking clear. At the very least, the oil film will be affected and parts may wear out faster than desired. To put the myth in proper context so it is correct: "By the time oil looks milky you're too late." (NOTE: The term "milky" should not be confused with other problems such as air entrainment and foaming which can also look "milky.")

MYTH No. 2 - "If the oil is bad then change it out and we can call it good and be done with it."

Not so. On half of all of the oil changes recommended by the oil analyzer, the oil had to be changed twice, and in several cases, three times before all of the contamination was eliminated; the contamination was that bad. Most of these multiple changes were on gearboxes. What was happening was the new oil was stirring up all of the sediment in the bottom of the gearboxes. The sediment did not flush out very easily, but some of it would go back into solution when new oil was poured on top. It also brought to attention the effectiveness of the actual method used in purging out the old oil; as a result, oil changes on equipment with really bad oil are done differently now.

MYTH No. 3 - "If the oil is full of water then we need to change all of it while the unit is down."

Not so. Tank agitator gearboxes are good candidates for doing a partial oil change on-the-fly. It has been done several times with very good results. The oil condition after the change on-the-fly wasn't perfect, but it brought the oil to within acceptable limits. This can probably be done on gearboxes with output speeds below 100 RPM if warranted.

MYTH No. 4 - "Clear looking oil is OK and certainly is not corrosive."

Not so. When water enters oil, (which we know we can't see most of the time), polar ions and molecules form. The continual breakdown of additives in the presence of water will give rise to the formation of acids. We can't see them, but these polar ions and molecules can support an electrical current. Therefore, they will affect the current that the oil analyzer uses in its detection mode, and the analyzer will see the problem before the human eye ever will.



  1. There are ways to store the sample test information incorrectly and not even be aware of the fact. The oil analysis software is somewhat user friendly for an experienced computer person, but to anyone who has never used a computer before, it is a nightmare.
  2. The hardware requires calibration checks religiously. It doesn't usually drift, but when it does it can ruin all of the tests unless detected while calibrating the system.
  3. The results can vary depending upon what part of the oil sample is analyzed versus proper mixing. To combat this problem the entire sample is mixed well and then placed into two separate containers, and then both are analyzed. This takes more time than the oil analyzer manufacturer will admit to.
  4. The solvent which is used to dilute the heavier oils for particle testing must be environmentally friendly. This way the sample bottles can be disposed of in the normal trash, and the solvent-laced oil sample can be disposed of in the normal oil reclaim tank. Inattention to this detail can be very serious in potential additional disposal costs and/or DEQ fines.
  5. Good ventilation is a must.
  6. The solvent must be clean so that it doesn't affect the particle detection by the hardware. It too must be religiously checked and calibrated for cleanliness prior to running any samples, otherwise, the oil sample may give false readings for contaminants if the solvent is dirty.
  7. The methodology used when taking samples from the machines is the Achilles heel, but this is true even if the sample was sent out to a lab. Poor sampling methods bring poor results.
  8. The computer term "garbage in, garbage out" comes to mind. Conscientious attention to detail is required in this program. The software is friendly enough to allow a person to make mistakes and not know it until it is too late.
  9. The ensuing frustration that goes along with inattention to any or all of the above mentioned points has the potential to make a person give up, or worse yet, indicate that "the analyzer doesn't work." This means that the person involved in such a program has to want to do oil analysis, otherwise the program will most assuredly fail.
  10. Doing an average of 21 samples in six hours may not be cost effective for some companies, as opposed to sending them out for analysis.



  1. Its like putting reading glasses on for the first time, there is no desire to go back. All of the guess work with regard to the effectiveness of the preventive maintenance program has been removed. It's an impartial judge as to the quality and timeliness of the Lubricator's work and the feedback is immediate. He can see for himself the results of any oil changes he has made, or lack thereof. It's a must in this "instant gratification" generation. No management/union issues here, just the facts.
  2. The Lubricator or maintenance engineer can react quicker to contamination problems by having an answer within an hour if required. This has happened several times already. No outside laboratory can equal that kind of response time.
  3. The oil analyzer assists in prioritizing any required oil changes. When manpower is short and the Lubricator might be called into doing mechanic work instead of lubrication work during a shutdown, there is a big difference between saying "we need to change this oil NOW, get help from somewhere else,"or, "the oil change can wait till next week."
  4. There is no longer any guesswork with regard to the effectiveness of partial oil changes. The Lubricator can see very quickly whether the oil is getting better and whether an additional partial change is warranted. With the oil analyzer it has already happened several times in the past where oil has been tested in the morning, a partial change made in the afternoon, and the results checked the first thing the next morning; a second oil change can be made by the time 24 hours has expired. No outside laboratory can facilitate that type of Lubricator response time.
  5. There are over 23,900 PM tasks for the plant Lubricator. The only way to survive and stay on top of contamination is to do only that work that is warranted. This is done by monitoring the condition of the equipment's oil using an in-house oil analyzer. The oil is changed only if it is bad, and not just because it is on a schedule. It takes time to change oil, whereas, taking a sample and analyzing it is a lot quicker.


Dirty Oil- Tank Agitator Gearbox

The graph to the left shows the oil test result on a tank agitator gearbox. The vertical axis shows the relative change in the condition of the oil, and it is a "dimensionless number." For example, a drop from 100 down to 80 is twice as significant as a drop from 100 down to 90. What we are trying to do here is to get a more qualitative "feel" for the condition of the oil, as opposed to visually looking at the oil and saying "yeah, it looks a little dirty." The horizontal axis is the time it takes for the analyzer to perform its analysis; in this case 500 seconds. The top line (blue) is the calibration curve which we developed for clean synthetic oil. Notice how it does not deviate on the vertical scale much in 500 seconds. Two things which the analyzer looks for is ferrous and non-ferrous contaminants. These contaminants are represented by the bottom line (green) and the middle line (orange) respectively. The further away the bottom line is from the top line at the end of 500 seconds, the more contaminated the oil. The further away the middle line is from the top line at the start of the test, the more corrosive is the oil. Each sharp deviation of the bottom curve which touches the bottom axis represents a ferrous particle 60 microns or larger hitting the analyzer sensor grid. Likewise, each sharp downward spike of the middle curve represents a non-ferrous particle hitting the grid. In this case, there were approximately 12 ferrous particles in the oil sample that were 60 microns in size, or larger. This was confirmed under a 100X microscope.

This oil was mildly corrosive and was relatively clean looking in appearance, excepting for the fact that there were too many chunks of contaminants. We did a partial change of oil on this gearbox because it couldn't be shut down at the time. The results are shown below.


Partially Changed Oil


Notice how the shape of the curves have changed. Did we solve our corrosion problem? Yes. If the middle (orange) line starts at the left with no separation from the top (blue) line, then the oil is not corrosive. The further away the orange line starts from the top blue line, the more corrosive the oil. When the bottom (green) line has no spikes dropping to the bottom of the graph then there are no particles greater than 60 microns in the oil. Notice that the green line has little deviation or small spikes. This means that there are very few suspended smaller particles in the oil. Did we clean the oil up totally? No. Did we improve it? Yes. Can we live with it? Yes. Any further work required? No, there are bigger fires to fight.

One important feature of this oil analyzer is the on board diagnostic software which the Lubricator uses. Looking at the bottom of each graph, notice that there is an evaluation for "Wear Condition" and for "Lube Condition." This evaluation can be customized from an extensive library of equipment standards that are provided by the oil analyzer manufacturer. In the first graph the wear condition is "Bad" and the lube condition is "Extreme." In the second graph , where we made a partial change of oil, the diagnosis of the wear condition hasn't improved , but the lube condition has improved significantly. Not shown here, the diagnostics give helpful hints where to look for the problem, and what to do to correct it.

So, did we eventually change this oil? Yes, when we had time. Did we get good results? Yes, we did. The graph below shows the oil condition after a full change.


After full oil change



Notice how both the green line and orange line are superimposed upon the reference line. This oil change was successful and it was done as time permitted as opposed to a crisis shut down.

If you're observant you might see that the analyzer says that the wear condition of the oil is still bad (bottom of graph). Unfortunately, we gave the analyzer the wrong standard to compare this oil to. We mistakenly told it to compare the sample to the rigid requirements of transformer oil. Those requirements are far more strict than for the standard we would have used. As mentioned earlier, it is easy to go down the wrong path with the software and not know it until it is too late.



Over the course of eight months it became quite clear that oil contamination did not follow a schedule. It followed a schedule similar to the frequency of heavy driving rains, leaking overhead equipment, and washing areas down. Therefore in retrospect, an oil PM schedule is just the equipment manufacturer's best guess at how long a machine will last before the oil degrades from natural causes, given perfect operating conditions. None of these perfect operating conditions exist in a heavy industrial environment.

Routinely changing the oil is a great preventive maintenance technique, but there is not enough manpower and time anymore to do all of the oil changes that the equipment manufacturers suggest or the maintenance engineers pile onto the PM schedule. Routinely analyzing oil is the best alternative. It can be done in-house, but if you're thinking of doing that, make sure the people that are going to be doing the analysis are very motivated. Otherwise, the program will fail because the favorite universal excuse used by anyone who doesn't know, or doesn't want to know, how to operate any device correctly is ....

"It doesn't work."


All Pictures and Text Copyright © 1999 - 2016 Contact Mr. Adler Adler Engineering LLC of Wyoming USA

Great care has been taken in the compilation of this article. However, no warranty, expressed or implied, including without limitation, warranties of merchantability or fitness for a particular purpose, are given in connection with this article or any article archived on this website. Although this information is believed to be accurate by the author, the author cannot guarantee favorable results will be obtained from the use of this article alone. This article is intended for use by persons at their sole discretion and risk. Since the conditions of product or material use are outside of the author's control, the author assumes no liability or obligation in connection with any use of this information. The author is not liable for special, indirect or consequential damages resulting from the use of this material.

No part of this article or any article archived in this website, or any part thereof, may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning or otherwise, without the prior written permission of the copyright holder R. H. Adler. Nothing contained in this article or any article archived in this website shall be construed as a grant of any right of manufacture, sale, use, or reproduction, in connection with any method, process, apparatus, product, composition, or system, whether or not covered by letters of patent, copyright, or trademark, and nothing contained in this article or any article archived in this website, shall be construed as a defense against any alleged infringement of letters of patent, copyright, or trademark, or as a defense against any liability for such infringement.