When a machine is designed, one of the major considerations that the engineer must make is the selection of proper rolling element bearings. Some of the factors considered in the selection of bearings are shaft speeds, loads (both axial and radial), the environment they will run in, how we will be able to re-lube, the duty cycle, etc. And with a few exceptions, the bearings selected for the machine should provide trouble-free life for 11 to 20 years or more. It is well known that under the right conditions, it’s not unusual for some bearings to last 50 years, especially inside gearboxes where we maintenance types rarely get to go! There may be a lesson in that sentence. Think about it.
The figure below illustrates some general rules of thumb that engineers are taught to apply.
L10 is a bearing company term which means, “The number of revolutions which 90% of a population of bearings are capable of enduring before the first signs of “Metal Fatigue” begin to be visible under a microscope on one of its rolling elements. (either races or balls)”. This is determined by the maximum amount of load a bearing at 33 and 1/3 Rpm for 1,000,000 revolutions will endure before any signs of fatigue shows up. The L10 life is a statistical projection. An L10 life of 1,000 hours means that 90 % of a group of identical bearings, operating in identical conditions, and under the same load, should last the projected L10 hours before that first sign of metal fatigue.
Bearing experts and Bearing Company engineers however, tell a different story about the “real world”. They say that while this is a useful and very conservative number for design purposes, the bearing SHOULD be able to actually run 500 percent or 5 times that long for a useful life.
HOWEVER, we are conversely told by the same guys, that in the real world, these same 90 percent of bearings never even make their lower level of performance or L10, hence the “Great Bearing Paradox” that is the result of the actions by us Maintenance Folks.
As we teach to our students, “Bearings never die of old age, because we, as maintenance supervisors, craftsmen, operators, lubricators, stores personnel, and others continue to invent all kinds of ingenious ways, to torture them, resulting in their pre-mature death long before they achieve their L10 Life Expectancy!”
The graph below shows how increasing load shortens L10 Life as radial load is increased to a point at which the bearing is compromised. Notice that if the life expectancy is 12 years at 280 pounds of load, look what happens if we double that load to 560 pounds. The expected life is reduced to just under 2 years, an “exponential” reduction in bearing life”.
Now, let’s consider our same bearing with a speed change. If we are running at 1,000 rpm and have a projected similar life expectancy of 12 years, what is the impact of doubling the speed to 2,000 rpm? As we can see, the bearing life is cut in half. At 3,000 rpms, it is cut in half again, but, the speed relationship doesn’t nearly degrade our life expectancy to the same degree as load does.
Some argue that the mechanism of failure is not overloading of the bearing material, but reduction in oil film thickness usually brought about by unbalance or misalignment of the machine. Obviously, oil film thickness must be compromised before any metal to metal contact and subsequent overloading of the bearing material can occur.
On the other hand, others argue that it is the actual overloading of the bearing material itself that is the primary cause of failure. At any rate, it is clear and agreed upon that excessive loads cause premature and too often unexpected bearing fatigue and eventual failure.
The machine designer carefully chooses bearings that should provide adequate life for the calculated speeds and loads. Our task is not only to detect impending bearing failure and carefully analyze the symptoms of failure, but to determine and act upon the root causes…using a combination of all tools that are available.
In doing just that we find that many additional defects and or assembly errors are introduced along the way which either add additional load, compromise the lubrication, or both. Assembly by construction and or maintenance often results in unnecessary looseness, overtightened bearings, bearing housing and shaft Fit and Tolerances that are just flat wrong (too tight or too loose), wrong length keys, pipe strain, incorrect placement of components on shafts like belt sheaves not being as close to the bearing housing as possible, wrong lubricants, components not balanced to G1.0 like impellers, missing or wrong coupling bolts, and many others. All of these factors of course add load and increase vibration which eventually and often dramatically shortens the bearing life, and just as often requires a lot of excess energy to drive the machine until failure.
On the last Pump Reliability Improvement Project we did for a customer who couldn’t run a high criticality pump for 6 weeks between scheduled outages we found all of the following problems related to this discussion. In no particular order, Defect 1) Mechanics said they achieved Precision Alignment Specs and “Of course we checked and eliminated Softfoot”. Upon careful re-check both were found to be untrue. Re-alignment and eliminating Softfoot resulted in a significant reduction in overall vibration as well as a 3% reduction in electrical power usage. Defect 2) Thrust Bearings being installed incorrectly, leaving the bearings with NO Preload and excess axial movement, which could have easily destroyed the impeller and or wear plate at a minimum. Fortunately this was found and eliminated as we double checked the rebuilt power end before installation. Defect 3) Impeller factory balanced at G1.0 was cut to a reduced dia. for the particular service pump specs and NOT rebalanced. Defect 4) with pumps in this area running hot some rocket scientist decided to increase the viscosity of the oil from 68 to 150 and at the same time (w/ bearings running 180/200F) add cooling water spraying on the outside of the cast iron bearing housings thus reducing diameter further and bearing clearance even more. Defect 5) Several o.e.m. spare shafts checked, including the one used in the rebuild, exceeded the Fit and Tolerance dimensions from both the pump manufacturer and the bearing manufacturer at the bearing fit surface on the shaft. Additionally, Defect 6) the pump running a mechanical seal was found to have a shaft 0.002 undersized where the sleeve fit the shaft AND the i.d. of the sleeve from a third party supplier was found to be 0.003” too large, resulting in a sleeve to shaft fit of 0.005 total looseness and they wonder why they can’t make mechanical seals run between outages. Defect 7) At 3600 rpm on a 2’ dia. shaft they were using a set-screwed coupling rather than a shrink fit. Defect 8) The same coupling was running out 0.012”. Subsequent checks revealed it bored off center in a 3 jaw chuck which every real machinist knows is the wrong way to do it! Defect 9) Using a noted pipe strain acceptance test, both the discharge piping and the suction piping (which failed of course) were causing the cast iron rotating power end and stainless casing to distort putting the bearing housings out of alignment to each other.
You cannot make this stuff up! We see it all the time when we do a deep dive into a failure. If it were not so criminal it would be funny.
In conclusion, If maintenance management would just realize the value of eliminating these defects AHEAD of starting and running rotating equipment, not only would uptime improve but we would not have to spend so much time and $$ inspecting, doing vibration checks, infrared, oil analysis, not to mention countless unnecessary replacements.
All of these are great and necessary tools but they simply help us mitigate failure by hopefully scheduling an expensive and highly repetitive failure to avoid only the unscheduled downtime.
In the future I hope to do articles on how these technologies in effect, help us hide the fact from plant management of the total cost of repairing pump after pump (and all other forms of rotating equipment by the way) for instance, every 2 to 3 yrs that should run 8 or 10 or, as an operating friend says, until the hydraulic components are worn to the point the operator says “ Make pump pump; Pump is not pumping the required amount!” That is the point we should overhaul pumps and seals and NOT Before.
We also hope to do an article soon on the related topic of the specific defect elimination expectations and standards that management should set, communicate, and hold the entire organization, especially mechanics and front line supervision, both operations and maintenance, responsible for seeing is carried out to insure equipment is ordered or rebuilt, stored, installed, adjusted, and run in a precision state.