Elam and his colleagues recently battled an outbreak of bearing failures that seemed determined to push those pathetic averages even higher. A consulting engineer and vibration analysis specialist, he works in global maintenance improvement at Eli Lilly and Co., the worlds 10th largest pharmaceutical manufacturer.

We were starting up a new drug manufacturing facility in Indianapolis, and had to install about 56 new fans, all grease lubricated, Elam told last months meeting of the Society of Tribologists & Lubrication Engineers, in Las Vegas. These fans are critical for maintaining the sterility of the filtered air which circulates in the facility, yet many began experiencing failures very quickly and inexplicably after startup. Losing a fan to a bearing failure could stop production, Elam noted, since a pure environment is crucial to drug manufacturing.

A typical case was Fan #107: belt driven, with a self-aligning kilowatt bearing, it was supposed to supply air (a filtered mix of fresh air from outdoors and recirculated air from the plant) back to the clean-room manufacturing area. Prior to startup, Fan #107 had been sitting idle for about a year. But soon after it was put into service, Elam began seeing very large vibration variances in the timewave form.

Vibration analysis showed that every time the ball passed over the outer race, it was generating noise, indicating trouble. This bearing unit was over a year old, but it was very new to service, so why were we seeing service problems?

The fans are two-horse-power units, with large motors that measure about 2 feet across. When working in the field, slow speeds tend to present a big challenge to vibration analysis, and normally thats where oil analysis can help, he told the meeting on May 17. Ive spent most of my career in the area of vibration analysis, but oil analysis has really come of age in just the past 10 to 15 years, and has been accepted by the vibration community. Grease analysis, on the other hand, is still in its infancy – and these were grease-lubricated bearings. Nevertheless, Elam and his co-authors, Bill Morgan in maintenance improvement services and Rendela Wenzel, the corporations lubrication team leader, decided to give it a try.

The Enigma of Grease

First step was to figure out how. Its far easier to obtain an oil sample, and most analysis labs have little experience with grease sampling. In the past, if you wanted to have your grease analyzed, the laboratory wanted a sample of 2 to 3 ounces, which is more than we could provide. Its almost equivalent to all the grease in a bearing, Elam pointed out.

The Lilly team turned to MRG Power Labs in York, Pa., which had developed a tool called the Grease Thief, a syringe-like extractor that draws only a very small amount of grease from a bearing. MRG required only 1 to 2 grams of grease, and could test one sample in a variety of ways to get quite a bit of data, including wear particle analysis.

Hopeful of finding the root cause of its losses, Lilly began sending grease samples from failed bearings for analysis. On occasion, if pressed for time, it would replace the bearing and then send the entire troubled unit to MRG, which itself extracted the sample.

We sent one bearing to MRG, and got a grease analysis report back showing evidence of very high wear, an unusually high amount of aluminum (almost 3,000 ppm), and high oxidation, Elam said. MRGs report on the greases condition also suggested it may have experienced cross-lubrication, perhaps before it went into service. That had been seen in other startups, where contractors who maintained the bearings prior to installation may have used all-purpose greases as if they were interchangeable, and mixed bentonite with polyurea greases, for example.

We also saw spalls in the bearings outer-face load zone, which accounted for the wear particles – but why was there so much aluminum? pondered Elam. That question became just one of many as the plant soon saw other problems, and dozens of fans began to fail.

Some of these failures were pinned on cross-contamination issues, where contractors had used the wrong grease. But evidence also began pointing to the bearing designs. The self-aligning bearings were intended to float in their races, but grease analysis showed high wear-particle counts and closer examination turned up a lot of fretting wear.

Another peculiarity was that often there were no vibration symptoms prior to seeing the evidence of fretting in the bearing. This led Elam and his team to declare that fretting was a root cause of the failures. Grease which should have helped prevent this did not seem to reach the bearing in an efficient way. The Zerk fitting where the grease was pumped in led to a long groove that circled almost completely around the pillow block before reaching the bearing itself, so the grease had to travel an over-long path to its destination.

Oddly, a lot of the failures occurred shortly after the six-month preventive maintenance was performed, Elam related, when fan drive belts were examined and tightened or other routine work was done. In some cases, we found that maintenance was just flat-out using the wrong grease, a failure to follow basic lubrication practices.

Searching for Roots

The root-cause analysis dug further back, into the plants history. This particular facility was originally slated for startup in 2008, but its equipment sat idle for one year waiting for final FDA approval for the drug it was intended to make. Thats not unusual, Elam noted. Ideally, a plant will come onstream just as the new drug gains the U.S. Food and Drug Administrations nod, so it can hit the market running. But if the approval is delayed, as happened in this case, the plant may stay under wraps – literally.

For about a year or so, the fan motors and bearings were stored by a contractor, perhaps in a warehouse but just as likely in a yard under tarps, Elam recalled. The contractor was responsible for regularly lubricating the units, but went out of business before the plant was finished; another contractor stepped in and completed the work.

Could one or the other contractor have used the wrong grease when the bearings were stored or installed? Lilly may never know for sure, Elam acknowledged. But he says long experience has shown that many untrained workers wrongly believe that all greases labeled all-purpose are interchangeable, so such errors are not uncommon.

In the end, Lilly changed out 200 bearings, and replaced them with units from another OEM. The new ones had far more rigid pillow blocks and tighter clearances between the bearings outer diameter and the pillow blocks inner diameter. And the Zerk fitting is very close to where the grease feeds the bearing. That seemed to solve the fretting issue.

Housekeeping also helped. We insist the grease not be interchanged, Elam said. We also label the grease fittings with the correct grease, and use dedicated guns for every grease type. So a grease gun for lithium grease cannot be used for anything else, such as polyurea, and vice versa.

Throughout the process, grease analysis seemed to closely correlate with the bearing vibration analysis. Grease analysis is far more expensive than oil analysis, however, and can run hundreds of dollars per sample, depending on the tests ordered. On the other hand, a failed $200 bearing can cost us a half-million dollars in downtime, the engineer pointed out.

We used 16 to 18 samples in all with MRG. Due to the cost its not a technology for every bearing issue, but we needed to find the root cause of these failures. A new drug, Lillys lifeblood, was riding on the outcome.

A Look at Lilly

Col. Eli Lilly, a U.S. Civil War veteran and pharmaceutical chemist, founded his company in Indianapolis in 1876, and it remains an independent maker of medicines. Lilly was one of the first to launch a bona fide pharmaceutical research program, in his companys first decade.

From there, the company saw a stream of historic milestones, among them the introduction of the worlds first commercially available insulin product (1923), the first mass-production of penicillin (1940s), and the first human health-care product created using recombinant DNA technology (the diabetes drug Humulin, in 1982).

Lilly today has around 39,000 employees worldwide (over 11,000 of them in Indianapolis), and manufacturing plants in 13 countries. Revenues in 2009 were $21.8 billion, and net income $4.3 billion. Like other pharmaceutical companies, its success depends on keeping its new-product pipeline full, especially as older medicines come off patent and turn generic as Prozac did 10 years ago. Lilly responded to that loss by launching nine products in five years, including best-sellers Cialis, for erectile dysfunction, and the pain and anti-anxiety drug Cymbalta.

Last year it spent $4.3 billion in research and development expenditures. It can take $800 million to $1.2 billion to discover and develop a new drug, the company says, and the average length of time from discovery to patient is 10 to 15 years. The payoff can be spectacular though: Eight of Lillys drugs had sales over $1 billion worldwide last year, led by the schizophrenia drug Zyprexa at $4.9 billion.

Meanwhile, Back at the Plant

In its plants today, Elam said, Lilly ranks equipment in three categories of criticality, and schedules lubrication, vibration checks and oil and grease sampling accordingly. The rankings are assigned by reliability engineers, with attention to Good Manufacturing Practices, as required of pharmaceutical operations.

Elam himself joined Lilly seven years ago, after a long career with Duke Energy. He credited Rendela Wenzels leadership, and that of her predecessors, for the lubrication teams success.

Weve been active in getting Lillys plants to meet the basics of good lubrication, and we even conduct lubrication audits, he said. We also focus on training, to elevate the role of the people who do lubrication. In lots of places, many people consider lubrication to be unskilled or low-ranking work. So we have a program to recognize the Outstanding Lubrication Specialist each year, awarding a plaque with their name on it, and getting their photo in the company newsletter with the top managers.

A final question arose from the STLE audience: What about those high levels of aluminum particles that MRG Power Labs spotted in the grease samples? Their source is still unclear, Elam replied. But the air-filter frames in the plant, it turned out, are made of aluminum, and we theorized that as the frames vibrated, they may have released wear particles.

Could these particles have been sucked into the bearing grease? Were not fully certain, but its our best answer.