Pulp and Paper Canada

Features Equipment & Systems
Using Bearings

Unlike many products on the market these days, bearings are fairly mature in that they do not see frequent introduction of new designs. That is not to say that the research and development, or develop...

March 1, 2007  By Pulp & Paper Canada

Unlike many products on the market these days, bearings are fairly mature in that they do not see frequent introduction of new designs. That is not to say that the research and development, or developments made in the application of bearings, is not ongoing. What will be discussed here are some of these developments such as the evolution of the bearing fatigue life equation and how the gap between actual bearing service life and calculated life is narrowing. Also one discussion topic that every bearing user will have to come across: bearing failure analysis. Can anything be learned from a failed bearing? How should a root cause failure analysis be carried out? And finally, some discussion on an often-overlooked topic regarding minimum load on a bearing.

More advanced life calculations

In many pulp and paper applications, the operating conditions for the bearing can be quite demanding. Not only are high speeds and loads influencing factors on bearing life, but also contamination and the effects of poor lubrication. It is important when selecting a bearing that all of these conditions be taken into account to best calculate a representable bearing life. To best do this, more advanced calculation methods must be used.


For rotating bearings, the L10 life equation has long since been used to measure the expected life of the bearing for given operating conditions. Despite the popularity of the L10 equation, there are some shortcomings, which can introduce discrepancies in the calculated life to that of the actual life. The L10 life equation is given as:

C p 106

L10h = (—-)* —-

P 60n


L10h = life in hours

C = dynamic capacity of the bearing

P = equivalent load on the bearing

p = calculation factor: 3/10 for roller bearing, 3 for ball bearings

n = rotation speed

After looking at the equation, it becomes apparent that the dynamic capacity, C, is the only variable relating to the bearing. The dynamic capacity is a calculated value based on the macro geometry of the bearing: ball or roller, number of rolling elements, size of the rolling elements, number of rows, and contact angle. The speed variable is simply for converting the life from number of revolutions to hours. Therefore key influencing factors, such as lubrication and contamination, are not taken into account in this life calculation.

In 2006, to factor the lubrication and contamination effects on bearing life, ISO adopted the SKF Life Theory. This life theory is given as:

C p 106

L10mh = askf (—-)* —-

P 60n


L10mh = life in hours

askf = SKF life modification factor

C = dynamic capacity of the bearing

P = equivalent load on the bearing

p = calculation factor: 3/10 for roller bearing, 3 for ball bearings

n = rotation speed

The askf factor takes into account the effects of lubrication and contamination and acts as a multiplier to the standard L10 equation, therefore making bearing life calculations more applicable to real conditions. In addition, the SKF Life Theory utilizes the fatigue load limit, whereby bearings operating with loading under this limit will not experience fatigue failure. This is best explained in an example.

A common bearing application in a paper machine is a dryer roll. Generally speaking, this is a demanding application, since steam being supplied via the hollow bearing journal causes a very poor lubrication condition for the bearing, and also contamination in the oil can be a persistent problem. Both of these situations can have a significant impact on bearing life.

The table below shows various cases in which the bearing in the dryer cylinder can be seen. For every case, the table will list the specific conditions, the temperature of the bearing, the contamination level (which is referred to as the eta-c value), the quality of the oil film (or the Kappa value), and the standard L10h life and the the L10mh life theory. To explain further; the eta-c value is an indication of the contamination level in the lubricant. The range is from 0 to 1, with 0 indicating complete contamination and 1 indicating extreme cleanliness typical of laboratory conditions. The Kappa value is an indication of the quality of the lubricant film and is the ratio of the operating viscosity to that of the required viscosity for complete separation. A kappa value of 1 or above indicates full separation, and less than 1 indicates that metal-to-metal contact will be occurring.

Case 1 is the existing application, while cases 2-4 represent condition changes made in the effort to improve the bearing performance. The effects on the oil film or contamination level resulting from these changes are shown in bold in the chart. Notice how the basic L10h life does not reflect any of these changes and the life remains unchanged at 93800 hours. This is not surprising since the life equation only factors in the bearing capacity, load applied and rotational speed, all which remain the same in all four of our cases. What is interesting is the SKF life, L10mh. Notice the life can vary dramatically, specifically in the case where the operating temperature of the bearing is reduced. This is achievable by using an insulated journal in the roll. This in turn causes the bearing temperature and therefore the lubrication temperature to drop, resulting in a higher operating viscosity. This dramatically improves the quality of the oil film and therefore results in much higher bearing lives.

Before more advanced life calculations were available, any changes that affected the operating conditions; lubrication, contamination, etc., were unaccounted for in the L10h life equation and the effects of these changes were most often validated by a “wait-and-see” approach, or by previous experience. Now the effects of any potential changes can be evaluated without having to utilize the trial-and-error approach. Further advancements are underway to take this even further to factor in more variables such as flexible housings, internal clearance and misalignment.

Root cause failure analysis

Undesired bearing failures do occur. However, there is a lot of information that can be taken from a bearing failure that can be used in the corrective action process. Although bearing life calculations are the key design tool used in selecting bearings, only 1% actually fail due to pure fatigue. The majority of bearing failures are related to a lubrication-related issue, with other contributing factors being contamination and improper mounting. What can be taken from this is that 95% of bearing failures can be either prevented or have their service life extended. One very important tool in doing this is root cause failure analysis.

When it comes to bearings, root cause failure analysis is generally viewed as a two-stage process: what is the root cause specific to the bearing, and what may have caused that specific bearing root cause. Understanding that this may sound confusing, this is best explained via an example: Imagine a failed bearing being looked at in order to determine the root cause failure in the hopes of potentially avoiding a similar failure in the future. After examining the bearing in a systematic approach by cleaning, disassembling, and investigating each component, it was noticed that the failure modes visible on the bearing indicated a breakdown in lubrication. Therefore the reason behind the bearing failure is inadequate lubrication. But, logically, the next question would be; what caused the lubrication breakdown? Many times this second question cannot be answered by looking at the bearing alone. It is by answering this second question that real corrective actions can be taken. Again, going back to our example, the next steps would be to look at the application as a system, the reason being that many times a bearing is a result of a system failure. The first question would be, what is the application and what are th
e operating conditions? In our example, say the bearing was taken from a dryer roll that failed prematurely. Were there any other bearing failures from the dryer system? No. Does one central lubrication system service the entire dryer? Yes. This already tells us some very valuable information in that only one bearing failed, but there is a common lubrication system for all the bearings. This could mean that there are more bearings potentially ready to fail, or it means there could be something unique to this particular bearing location. The next step would be to investigate more closely the surrounding environment: housing, steam joint, piping, etc. Again in our example it might be found that the piping was damaged, reducing the flow to the housing, thus starving the bearing of lubrication.

What is important to understand is that by just looking at a bearing alone, only a root cause for the bearing failure is possible. From this, possible factors leading up to that root cause can be suggested, but these can be further narrowed down by understanding the application and other influencing factors such as: process, maintenance schedules, looking at surrounding equipment, etc. The most effective root cause analysis is that which is carried out by a team. For example, this team could involve a bearing specialist, a process engineer, and maintenance person. By understanding what normally should happen in the application and by looking for evidence for changes to this norm, root cause can be determined.

A common question from bearing users is, “how much life is left in the bearing?” This can be associated with a bearing failure, or can be related to a bearing taken out of service without any signs of damage. Up until now, the answer was mostly based on a calculated life minus the amount of hours already run on the bearing. Now technology exists to evaluate the residual life in the bearing. By X-ray diffraction it is possible to analyze the residual stress and the micro structural decay in the failed bearing. This analysis gives indications of what loads the bearings have been subjected to. The-sub surface material response will be measured and compared to the predicted pressures and related to bearing fatigue life. Below is a photo of the apparatus used in X-ray diffraction:

Minimum load

It is easy to understand that a heavily loaded bearing should experience a shorter bearing life than that of a moderately loaded bearing, especially after understanding the life equations as described previously. However, there is a point where the bearing will be too lightly loaded for it to perform properly. Many times bearing application engineers have received inquiries from customers, asking why their bearing had failed so quickly. Upon being asked how much load the bearing was experiencing, the customer usually replies the bearing had hardly any load at all, and, based on the life equation, it should have lasted forever. They are then even more surprised to hear that this light load is exactly the reason the bearing failed.

To ensure proper bearing operation, there must be a minimum amount of load on the bearing. To understand this further, it is necessary to understand the concept of the bearing load zone. Using the example of a bearing with inner ring rotation, unidirectional inner ring load, the load will be supported on a portion of the outer ring. This portion or section is known as the load zone, and is analogous to that of a piece of pie. The size of the load zone depends upon the magnitude of load applied. The larger the applied load, the bigger the load zone or piece of pie; the smaller the applied load, the smaller the load zone.

The total applied load is supported by the rolling elements in the load zone. Therefore, any rolling element within this load zone is in a state of pure rolling along the raceway. As the rolling element leaves the load zone, the rolling element is unloaded and may not necessarily roll along the raceway, it may “float” along, suspended in the lubricant between the two raceways. As the rolling element re-enters the load zone, it returns to pure rolling.

In a bearing that has enough load (enough meaning it meets the minimum load requirement) the contact pressure between the rolling element and the raceway gradually increases as you enter the load zone to a maximum at the middle of the load zone, then gradually decreases as it leaves the load zone. This is important since the rolling element would not see a sudden change of angular velocity. If the bearing is operating in an under-loaded condition, the load zone is very narrow. What this means is that the contact pressure between the rolling element and the raceways suddenly changes from virtually nothing to maximum immediately as it enters the load zone, and is quickly released upon leaving the load zone. The results in very sudden changes to the rolling element’s angular velocity. This sudden change in angular velocity causes the rolling element to skid as it quickly accelerates to pure rolling in the load zone.

This is analogous to that of an airplane landing on the runway: on approach, the landing gear is down and the wheels are not moving, which is similar to the rolling elements. As the plane touches down on the runway, which is equivalent to the bearing load zone, the wheels must very quickly accelerate to match the speed of the plane. As it touches down, it smears the runway with rubber. This is exactly what happens in the bearing; the rolling element breaks though the lubrication film and smears the raceway.

There are various factors that affect the effect of minimum load situations in rolling bearings. The biggest factor is the load. A large enough load zone must be present to prevent sudden changes in contact pressure. The faster the bearing speed, the more drastic the angular acceleration of the rolling element. Other factors include viscosity and bearing clearance.

There are some measures that can be taken to prevent bearing skidding, or reduce the effects of the bearing skidding. As stated earlier, the single biggest factor is the load. Therefore, selecting the proper bearing size is crucial to ensuring the best load distribution is achieved. There are however, cases where the bearing size is predetermined by other factors: existing housing size or shaft size and the bearing will be lightly loaded. Some methods to minimize the effects of the minimum load include:

* Adjust the internal clearance – by reducing the internal clearance the load zone should increase, allowing for a smoother transition of the rolling elements in and out of the loaded area. Care should be taken to not reduce the clearance too much.

* Reduce the rotational speed of the bearing – the faster the rotational speed, the higher the minimum load requirement. This is primarily due to the greater rotation speed differential between rolling elements in the unloaded zone compared to those in the loaded zone. This increased speed differential increases the amount of angular acceleration upon entering the loaded zone, thus increasing the likelihood of roller skidding.

* Use oil lubrication in place of grease – the thickener of the grease can have a slowing effect on the rolling elements outside of the load zone, potentially leading to increased angular acceleration as the rolling element comes into the load zone.

* Special coatings – these coatings don’t prevent the skidding but rather reduce or eliminate the damage associated from the skidding.

What is next?

As most pulp and paper mills will focus on increasing output though increased speed and reliability, rolling bearings will be called upon to support such improvement programs. It is doubtful that the bearing industry will see the addition of very many new bearing designs, if at all, but rather the refinement of existing bearing designs. This will most likely focus on areas such as materials and steel cleanliness, improved manufacturing methods
and quality control, and special products such as coatings or ceramic technologies.

Parallel to this, bearing manufacturers will continue to better understand applications and continually develop engineering support in the areas of life theory and application engineering to more accurately understand the use of bearings.

John Melanson, P. Eng. is an Engineering Manager with SKF Canada Limited.

Differences in Calculated life of a 22244 CCK C4/W33 in a Drying Cylinder with VG220 oil

Case Conditions Temp(C) Contam.(hc) Oil Film(k) L10h(hrs) L10mh(hrs)
1 Existing Application 110 0.5 0.6 93800 59500
2 Filter to improve Contamination 110 0.74 0.6 93800 84800
3 Using a heavier oil (VG320) 110 0.5 0.7 93800 110000
4 Reduce Temperature 90 0.5 1.0 93800 257000

Print this page


Stories continue below