Overhung Fan assemblies Which bearing should be fixed and floating?


I would like to get some input on were a floating bearing should be installed in a overhung fan assembly. I have heard that the floating bearing should be on the fan side and fixed bearing on the drive/pulley side. I have also heard the opposite. Is there a standard? Or does each application differ? This would be for a belt driven fan.

Original Post
Let's assume a direct drive. The NDE bearing ( fan side) is loaded radially more then that of DE ( since impellor weight is closer ). It is better for a bearing to have axial load developed during operation applied on the bearing which has high radial.In order to satisfy that the NDE one has to be fixed.

Just my opinion...


I believe the presenter was emphasising that fans, overhung in particular, can have low radial loads and relatively high axial loads. The right load ratio is essential to ensure correct operation of the bearing, Fa/Fr < 0.3 (approx). - I am not sure of the presenter's source for this as I have seen SKF documentation indicating

For Spherical Roller Bearings Fa/Fr < 1 and,
For Self Aligning Ball Bearings Fa/Fr < 0.25.

This can best be achieved by ensuring that the axial load is taken by the bearing carrying the higher radial load and therefore this is the bearing to be locked. If this ratio cannot be achieved with either bearing then a redesign is recommended.
Gentlemen, I am the author of the presentation referred to earlier. The slides from the presentation are not really a stand alone document, so for a more detailed description of the issues and solutions please refer to the 2 documents that I have attached. These are "Maintenance Skills and Practices Flyers" that we have produced on this topic. They should clarify things for you about which bearing to lock, but if you have any questions after reading them, please ask.
A word or two on Carb bearings. These are often promoted as the ultimate solution for overhung fan bearing problems. The reality is that if you have spherical roller bearings and an overhung fan that has catastrophic bearing failures, then Carb bearings are probably not the answer. As I explain in the Flyers, the problem is often with the FIXED bearing not having enough radial load. The floating bearing is not usually the problem. But Carb bearings cannot be used as a fixed bearing, so they don't help in this situation. And using them as the floating bearing doesn't achieve anything - apart from complicating setup and parts compatibility issues. There is, however, one situation I'm aware of where a Carb bearing does help. I know of a manufacturer who had a fan that rapidly went out of balance, and they were unable to rebalance the fan often enough. This resulted in the plummer blocks being pounded to the point where the bearings began to turn in the housing, and this ultimately led to bearing failure. In this situation the Carb bearing can help because the bearing outer race is actually clamped in the plummer block (standard plummer block/bearing arrangement has clearance between bearing and housing)and any axial float is accomodated within the Carb bearing.
My final piece of advice is that if you are having failures, then do a root cause analysis on your bearing failures. Only then can you determine what the problem is that needs to be fixed! At that point you may find that these Flyers offer a solution.
Rod Bennett.



Thanks for the attachments. I was advocating this concept all along for a good chunk of time ( see my brief post above ) and glad to have an ally.

There is another related issue here though. The DE bearing in a direct drive is not meeting most of the time minimum load requirements. This leads to overheating. Not changing the outlined approach it appears to me that the DE bearing has to be of a smaller size then NDE.

This is an excerpt from an SKF paper in Plant Services on this topic:

One cause of fan failure is confusing the fixed and floating bearing positions. A typical fan has two bearings supporting the shaft. In most cases, the bearing exposed to the highest radial load should be fixed, or axially held, within the housing. Radial bearings generally perform better when the axial-to-radial load ratio is low.

Also, the axial force needed to float a bearing is lower when the radial load is lower (axial force to move bearing = coefficient of sliding friction x radial load) (or Fa = µ x Fr). The fan’s other bearing should be allowed to float internally or within the housing bore, thus accommodating shaft expansion and contraction.

In most fan applications, the bearing closest to the motor or belt drive has the highest radial load and should be the fixed bearing.
I have recently learned that one particular belt driven over hung fan manufacture requests the fix bearing be at the fan wheel side and the floating bearing at the pulley side. The answer that I received was that the load on the bearings were too low, thus why they specify the fixed at wheel and floater at pulley. We just replaced bearings on this brand fan, and have installed them per the drawing/mfg specs. Time will tell if they last longer than the previous bearings.
Being just a bit of the devil's advocate, assuming the amount of axial float within the "floating" bearing is reasonably limited, and that the fan wheel-to-housing clearance is adequate, what happens if you allow both bearings to float? For most of the bearings I see on overhung fans, fixed vs. floating is simply a matter of inserting or removing the "rings" - otherwise the bearings are usually identical.
Originally posted by rustythevibeguy:
what happens if you allow both bearings to float? .

IMO it will be not healthy for the machine. After all for proper machine operation one wants shaft's position fixed but the bearing can potentially move inside the pillow block housing within a 0.25" range. Obviously one can't make both bearings fixed due to thermal expansion of the shaft. Therefore just one bearing is fixed with maximum (undesirable) shaft movement in axial direction limited to thermal growth.
I have a customer that has two small fans (18" dia.?), with bearings maybe 8 or 10" apart, running about 1900 RPM, that have both bearings float. They came from manufacturer that way, and have been that way for years. I had them fix one of the bearings on one of them a few months back (fixed the IB, sheave side)and then balanced it and ran. We really couldn't tell that much difference in the running vibration of them when looking at the trends. They both run fairly high in 1X, with two floats or 1 fixed, 1 float. Confused

Originally posted by Rod Bennett:
As I explain in the Flyers, the problem is often with the FIXED bearing not having enough radial load. The floating bearing is not usually the problem.

It is interesting that for direct driven overhung fans with both DE and NDE being spherical roller bearings some OEMs make DE bearing FIXED even if it is radially loaded only 20-25% of that of the NDE bearing. It is possible that in these cases NDE still has sufficient radial load to deal with axial loading and at the same time provide for proper coupling operation.

I will not go through theory. We have hot nitrogen gas handling fan temp 200 degree centigrade. Previously in this fan at NDE side two nos cylindrical roller brg NU332 and at DE side two nos anguler contact ball brg 7330B which was locked. So more radial load at NDE but DE side is locked now we have modified the brg combination with one no 22332 Double Row Spherical brg at NDE which is floating and 22330 one no at DE side which is locked and running smoothly.
Now the second example same fan in other section which is handling same gas with same temp but different brg combination that is three nos brg at NDE one is NU230 and 2nos 7230B anguler contact ball brg, out of this two ball brg one is radially free to take Axial load and in DE side NU226. Here NDE side is locked for same load and same appplication.
So my opinion is any side can be locked.
Regarding the puzzling OEM recommendation, the load ratio is quite important for spherical rollers. It's possible they moved the fixed bearing to the fan side improve the load ratio.

I know from hard personal experience that sphericals and thrust load must be carefully managed. The ability of a spherical to handle high thrust is very dependent on how smoothly it starts and stops and whether there are changes in either the radial or thrust load during operation.

I have same type of overhung fan. In one design Locating bearing is at Fan side and in another design it is Motor side I mean DE side. Both the fans are having same application and the handling temp is 180 degree centigrade. Any side can be located.


I have seen some fans with the DE bearing floating and some with the NDE floating, but I cannot ever recall seeing an overhung pump with the DE bearing floating --- not to say they don't exist. Most center hung pumps have the NDE bearing fixed, but all the overhung pumps I've seen have the DE bearing fixed to the shaft and housing and most allow the whole assembly to be adjusted for impeller clearance, but the DE bearing is still fixed once the impeller is adjusted and the NDE bearing sets in an open bore with the ability to float (see the attached picture of a Goulds Pump).

John J


Files (1)
Originally posted by Danny Harvey:
Maybe a very, very close clearance between the fan rotor and housing?

That is the reason I was given by a fan manufacturer of a pressure fan that was about 3" wide and ran around 3000RPM. They said they did not want it to contact the housing. I didn't either.

I had one client replace some worn bearings recently and they installed two floating bearings in a direct coupled overhung fan with a Falk grid coupling.

They realized and corrected their mistake before I was on site again. From their description it was quite evident that something was wrong but given the process, it may have had to run for a while after restart.

What WASN'T evident was the fact that the motor bearing cage was broken by the axial hammering it took when the free-floating fan shaft collided with the fixed motor shaft.

So it required a 2nd unscheduled shutdown after our next regular visit revealed a great big anglefish at cage frequency. (Good catch, Climb! Wink)

I've also seen stainless clad shafting lose its cladding and huge screw conveyor pipes break because of two fixed bearings.

So I can add one thing to the discussion:

You almost always need a fixed and floating bearing somewhere. Where should be determined by the designer.
I believe it's Fan Specific. Generally the DE Bearing is Fixed and the NDE Bearing is Floater.
On belt driven fans with High Side Load and Light Wheel, they will Fix NDE Bearing.
On Direct Drive (arr.8) with a Light Wheel, they will Fix the NDE and use a ball bearing in DE
(22XXEKC3) also on Arr. 8, a couple of manufacturer's drop down 2 sizes on the DE Bearing to keep it Loaded. ie. 2 15/16" wheel end bearing and 2 7/16" drive end bearing.
They are trying to prevent "skidding" so if you have any doubt and the fan still has it's ID tag
try to get original drawing. Frequent failures bearings should be analyzed in case site conditions differ from original fan engineering specs provided during procurement. When you say you have an "Overhung Fan". Weight of Wheel, Axial Load, Belt Driven or Direct Drive takes the same fan and produces different scenarios.
Class 1--Class 2--Class3---the wheel is dimensionally the same, say 27" OD x 10" wide
The difference is the thickness of the material.
Class 1 lightest--HVAC Class 3 thicker material for Process Fans. That's why I say-Fan Specific
In one of our overhung radial fans, the radial load near motor/coupling end is lesser than the radial load near fan wheel. Frequent high vibrations and high bearing temperature were reported. I had seen above case studies and refered to our JV at Italy and informed them whether we can make coupling side bearing as floating and fan wheel side bearing as fixed. But they have refered some SKF selections and suggested us to keep coupling side bearing as fixed type only. Is it right?

Add Reply