7/28/2019 12FTM02_Soendgen_17122

1/11

12FTM02

AGMA Technical Paper

Power Loss and AxialLoad Carrying Capacityof Radial CylindricalRoller BearingsBy S. Sndgen and W. Predki,ZOLLERN GetriebetechnikDorsten GmbH

7/28/2019 12FTM02_Soendgen_17122

2/11

Power Loss and Axial Load Carrying Capacity of RadialCylindrical Roller Bearings

Dr.--Ing. Simon Sndgen and Prof. Dr.--Ing. Wolfgang Predki, ZOLLERN

Getriebetechnik Dorsten GmbH

[The statements and opinions contained herein are those of the author and should not be construed as anofficial action or opinion of the American Gear Manufacturers Association.]

Abstract

The application of cylindrical roller bearings (CRB) is widespread in mechanical engineering. CRB can carrycomparatively high loads and are usable in high speed ranges. These bearings have been proven to bevariably applicable and economic. With lipped inner and outer rings CRB permit the transmission of axialloads in additionto radial loads. The axialload is induced on the lip of the inner orthe outer ring and transferred

by the roller end face contacts to the opposing lip. In comparison to an only radially loaded bearing, there areadditional friction losses in the contact between the lip and the roller ends as a result of sliding.

The limiting factors for the permissible axial load are high temperatures, which can cause smearing andseizing, lip fracture, fatigue failure and wear. In consequence of the axial loading, the stresses in the contactbetween the roller and the raceway rise and the fatigue durability of the bearing is reduced.

At high speeds the permissible thrust load is dominantly l imited by high temperatures. At low speeds thelimiting factors are lip fracture and wear.

Within the examination an extensive test program with different bearing geometries is carried out. Therebythe decisive measure is the friction torque of the bearings.

The friction torque of a thrust loaded radial cylindrical roller bearing is mainly dependent on the parametersspeed, load, size and design of the bearing.

An analytical simulation model, which has been developed at the institute, allows calculating the lubricationconditions, the stresses within the lip - roller contact and the axial load dependent friction torque.

The intention of the study is to enlarge the application range of radial cylindrical roller bearings by means of amore precise determination of the thrust load capacity and to allow more economic designs.

Copyright 2012

American Gear Manufacturers Association1001 N. Fairfax Street, Suite 500

Alexandria, Virginia 22314

October 2012

ISBN: 978--1--61481--033--9

7/28/2019 12FTM02_Soendgen_17122

3/11

3 12FTM02

Power Loss and Axial Load Carrying Capacity ofRadial Cylindrical Roller Bearings

Dr.--Ing. Simon Sndgen and Prof. Dr.--Ing. Wolfgang Predki, ZOLLERNGetriebetechnik Dorsten GmbH

Introduction

CRB have a wide application range and high significance in drive engineering. Due to the line contact CRBhave a greater radial load capacity than other rolling bearings of the same size. The possible application athigh speed and an advantageous friction behavior are additional benefits of radial CRB [1]. Several designs,which vary regardingthequantity andarrangementof thelips can be distinguished. The lips allow theapplica-tion of axial loads in one (supporting bearing) or two directions (locating bearing). When axial forces areapplied, they are transferred from the lip of the inner ring to the roller end faces of the roller and from there tothe lip of the outer ring for example. Alternating axial loads can be supported by a bearing of the NJ--type incombination with an angle ring HJ or by a bearing of the NUP--type with a loose lip [2].

Despite the fact that combinedloaded CRB are employed for more than100 years, there is no standard for thecalculation of the axial load dependent friction torque and of the axial load capacity until now [3]. There areequations given in the relevant literature and in the bearing catalogues which allow the calculation of the axialload dependent friction torque. These equations are mostly based on empiric approaches and the results ofthe different equations vary significantly.

Lubenow [4] performed extensive measurements on combined loaded CRB. Figure 1 shows the results fromtheequationsof Schaeffler [5], SKF[6], Braendlein [7] andthe simulation according to Lubenow as well as themeasurement results accordingto [4]. Theillustrated results arevalid fora runin bearing of thetype NJ 210atan operation temperature of 70C and lubrication with a mineral oil of the viscosity class ISO VG 220.

Figure 1. Axial load dependent friction torque T2 according to different calculations methods andmeasurement results

7/28/2019 12FTM02_Soendgen_17122

4/11

4 12FTM02

The radial load amounts to Fr = 5kN whereas the axial load Fa = 4 kN is chosen. The axial load dependentfriction torque T2 is illustrated in dependency of the inner ring speed n.

Figure 1 shows significant deviations between the different determined axial load dependent friction torques.Particularly at low speeds not only the results from the equations of the bearing manufacturers and ofBraendlein vary from the measurement results, but also the simulation results according to [4].

The friction torque of thrust loaded radial cylindrical roller bearings is an important parameter for thedetermination of theaxial load carrying capacity at mediumto high speeds. The permissible thrustload canbedetermined by means of a heat balance [4]. At low speeds the frictional losses decrease and therefore thetheoretically permissible thrust load strongly increases. The bearing manufacturers therefore additionallygive a limitingratio of axial to radial load and a limiting axial load in dependency of the sizeand the design ofthebearing.

However, these details are only valid for the bearings of one manufacturer and considerable safety marginsare included.

Testing program

For this investigation [10] five different designs of CRB in three sizes areconsidered. For theevaluation of the

size effect, bearings with the bearing diameters d= 30, 50 and 80 mm are chosen. Rolling bearings areproduced in different series. These series feature varying load ratings in dependence of the outer diameter,quantity of rolling elements and the bearing width. Besides the bearing series 22, the series 23 is examined.Bearings of theNJ--design normally provide friction optimizedlipgeometries. The lip is notorthographic to theraceway but inclined with the lip angle. Furthermore the lip can be crowned. The loose lip of a NUP--designand an angle ring in contradiction do not feature a friction optimized geometry. The evaluation of these shallshow the influence of the lip geometry on the friction behavior of axial loadedCRB. Full complement bearingsemploy the maximum possible number of rolling elements, whereby the radial load is shared by a greaternumber of contacts which leads to greater load rating. These bearings can be used as a supporting bearinglike a NJ--design because they also feature three lips.

The lubricant has a large influence on the friction behavior and therefore on the permissible thrust load of

radial cylindrical roller bearings. Within this study a mineral oil and a polyglycol of the viscosity class ISO VG220 are tested. Table 1 gives a summary test program.

Table 1. Examined bearings and lubricants

Bore reference number 06 10 16

Bore diameter d= 30 mm d= 50 mm d= 80 mm

Variant

NJ2206 E.TVP2 NJ2210 E.TVP2 NJ2216 E.TVP2N2306 E.TVP2 NJ2310 E.TVP2 -- --

NUP2206 E.TVP2 NUP2210 E.TVP2 ----NJ2206 E.TVP2

+ HJ206NJ2210 E.TVP2

+ HJ210NJ2216 E.TVP2

+HJ216

SL182206 SL182210 ----Mineral oil Aral Degol BG 220

Polyglycol Tribol 800/220

7/28/2019 12FTM02_Soendgen_17122

5/11

5 12FTM02

All bearings feature normal clearance. The cage material is Polyamide 66. The friction torque measurementsare performed at different combinations of speed and load. The gaps between the different speeds in the lowspeed range are chosen smaller than at high speeds because the lubrication film thickness in the lip--roller--contact grows with speed. With rising lubrication film thickness the rate of mixed lubrication decreases andthe hydrodynamic friction increases. The focus of this study is on the operation with mixed friction, for whichreason the test speeds are chosen according to Table 2.

The tests at constant speeds are supplemented by run--ups. The test program is completed with run--in andwear experiments.

Friction torque measurements on rolling bearings

For the assessment of the behavior of combined loaded CRB as well as of bearings in general, the globalmeasure friction torqueis of decisive relevance. On the onehand theresulting bearing temperature is directlydependent on the friction and on the other hand conclusions to the lubrication condition can be drawn. In thepast years extensive measurements of bearing friction torques have been performed at the Chair ofMechanical Components and Power Transmission, on different test rigs.

Lubenow [4] as well as Koryciak [8] used devices which allowed the measurement of the friction torque on theouter ring of the test bearing.

For this study a new test rig has been designed, which allows the measurement of the friction losses of mul-tiple bearings with a torque measurement shaft. The detection of the friction torque is effected on the innerring. Figure 2 shows that the result is affected by the measuring position due to the splashing of the rollingelements and the cage in the lubricant.

Whilst the outer ring is standing still, the inner ring is driven with the necessary driving torque Tdrive. Thefriction of thewhole bearing hasto be overcome andtherefore thedriving torqueTdrive complies with theabso-lute value of themeasured friction torqueon theinner ring TIR. Due to thesplashingof therolling elements andthe cage the drag torque Tdrag results on the cage which is retained in the lubricant. The measured frictiontorque therefore is reduced by the amount of the drag torque. In the following the influence of the differentmeasuring techniques on the results is examined. With the help of a modified test rig, the influence of themeasuringposition dependenton speed andoil level is investigated. The testing bearings areapplied with the

minimum radial load according to [6]. Whilst operating with minimum lubrication the drag losses vanish, bothmeasuring positions should show the same results. Dipping the shaft into the lubricants leads to additionaldrag losses whichwould influencethe results. Thereforethe maximum oillevelis limited to theinnerring of thetest bearings. Between these extremes the levels lowest rolling element half in oil and lowest rollingelement in oil are defined. Figure 3 shows the results for the minimum oil level.

Table 2. Test speeds

Inner ring speed, n, rpm 20, 50 100, 500, 1000, 3000, 5000

Figure 2. Measured friction torques of a bearing

7/28/2019 12FTM02_Soendgen_17122

6/11

6 12FTM02

Results for minimum lubrication

Figure 3. Friction torque at minimum lubrication

Figure 3 shows the total friction torque in dependence of the inner ring speed. The measuring results arecompared with the calculation results according to SKF [6]. The deviation between the measured torques onthe inner and outer ring are within the measurement precision. This approves the assumption that with disap-pearance of the drag losses the measuring position has no influence on the results. Furthermore at lowspeeds the results fit the calculations according to SKF very well. At medium and high speeds the calculatedvalues are below the measured ones. Figure 4 summarizes the results for the maximum oil level. In additionto the SKF results, values calculated with the method of Koryciak [8] are visualized.

Figure 4 shows that with higher oil levels a rising difference of the measured torques can be observed from aninner ring speed of 1000rpm upwards. The measuring position on the outer ring furthermore shows a goodcorrelation to the results according to Koryciak. Koryciak performed his examinations on test rigs whichdetect the friction torque on the outer rings of the test bearings. The conclusion is, that with rising oil leveldiffering friction torques are to be observed on inner and outer rings. Furthermore theresults show deviationsin the range of 100 Nmm for the examined bearing. The measurements within this study are performed at

considerably high radial and axial loads, which lead to much higher frictions levels than observed in thischapter. In the range of high friction levels, greater deviations have to be taken into account, which easily canamount to several 100 Nmm. Considering this the influence of the measuring position on the results can beneglected.

Results for oil level up to inner ring

Figure 4. Friction torque at oil level Up to inner ring

7/28/2019 12FTM02_Soendgen_17122

7/11

7 12FTM02

Influence of the bearing size

For theexamination of thesize effect CRB with 30,50 and80 mm bore diameter aretested. Table 3 showstherelevant data of the NJ22--design.

Driven at constant speed, a rising bearing diameter effects a higher sum of velocities in the lip--roller contact.At rising sum of velocities the lubricant film thickness rises as well, which means that the sum of velocities has

great influence on the friction and wear behavior of the contact. At low speeds the lip--roller contact ischaracterized by mixed lubrication. While speeding up the portion of hydrodynamic lubrication rises. For thisreason there is a minimum at a certain speed range. Above the speed range with minimum friction the draglossesgrowbigger. Figure 5 illustratestheseeffects. The loadindependent losses are represented by T0,theradial load dependent friction torque is T1 and the axial load dependent friction torque is T2.

Caused by the differing sums of velocities, in connection with rising bearing diameters, thefriction minimum isremoved to lower speeds with rising bearing diameter. Concurrently, the axial load dependent friction torquedecreases, due to higher lubrication film thickness, with rising diameter and constant speed and load. Theaxial load dependent friction torque is described as the product of the normal force in the liproller contact, afriction coefficient andthe lever to thebearing axis. The lever grows with thebearing diameter, for which reas-on the friction torque grows as well. Figure 6 shows the result of the different effects at constant speed andaxial load.

Dueto thedescribedeffects thesmallest bearing shows thehighest axial load dependent frictiontorqueon thetest rig. For themediumsize bearing thelowest axial load dependent friction torque canbe observed whereasthe largest bearing has friction losses which range exactly between the other ones. Figure 7 shows theproportions.

Table 3. Bearing geometry NJ2206, NJ 2210 and NJ2216

NJ2206 NJ2210 NJ2216

Bore diameter, d, mm 30 50 80

Outer diameter, D, mm 62 90 140

Bearing width, B, mm 20 23 33

Quantity of rolling elements, n 13 16 18Rolling element diameter, dW, mm 9 11 16

Rolling element width, lW, mm 13 15 24

Static load rating, C0, kN 50 88 245

Figure 5. Distribution of the friction torque dependent on the inner ring speed

7/28/2019 12FTM02_Soendgen_17122

8/11

8 12FTM02

Figure 6. Friction coefficient and lever of the friction force on the lip of the inner ring

Figure 7. Axial load dependent friction torque T2 in dependent on speed and bearing size

For the interpretation of Figure 7, it has to be considered that the illustrated results are only valid for the axialload dependent friction torque and that speed, temperatures and axial loads are held constant.

Influence of the bearing design

The examined bearing designs are already introduced earlier. Macroscopically the different test bearings canbe distinguished by thequantity andconfiguration of their lips. Additionally full complement CRBand bearingsof the heavy series NJ23 are investigated. Table 4 shows the relevant data of the examined bearingdesigns.

As to be seen in Table 4, the full complement bearing and the NJ2306 obviously differ from the remainingbearings. In this connection especially the quantity of roller elements and their length are to be highlighted.Figure 8 shows measured total friction torques of the different bearing types determined in run--in tests. Thetest speed amounts to n = 500 rpm. The bearings are lubricated with mineral oil.

7/28/2019 12FTM02_Soendgen_17122

9/11

9 12FTM02

Table 4. Geometry of tested bearings with 30mm bore diameter

NJ2206 NUP2206NJ2206+HJ206

NJ2306 SL182206

Bore diameter, d, mm 30

Outer diameter, D, mm 62 72 62

Bearing width, B, mm 20 27 20Number of rolling elements, n 13 12 16

Rolling element diameter, dW, mm 9 11 9

Rolling element length, lW, mm 13 18 14

Number of lips on the inner ring 1 2 2 1 2

Number of lips on the outer ring 2 1

Figure 8. Run--in measurements

The bearing loads were chosenaccordingto their static load ratingC0, forwhich reasonthe total friction torqueof the variants NJ2306 and SL182206 exceed the ones of the remaining bearing variants. In the case of thefull complement bearing the roller elements contact. This contact induces higher frictional losses due to theirrotation in opposing directions.

Figure 8 shows a minor run--in of the NJ2206 during the first 50 operating hours. The reduction of the frictionamounts to approximately 500 Nmm and is considerably lower than observed in earlier examinations. This

observation canbe explainedby lower roughness beforethe start of thetest. As already mentioned, thelooselip and the angle ring do not feature an optimized lip geometry. Furthermore the roughness is higher than thatof the NJ bearing lip. For these reasons these variants show a more distinct run--in effect. The combination ofa NJ2206 and an angle ring reaches a similar friction level to a thrust loaded NJ2206 without an angle ring.The NUP--bearing does not reach thefull run--in state as canbe seen in Figure 8. Tests with the largerbearingNUP2210 show that a complete run--in bearing with a loose lip can have a similar friction torque like a NJ--bearing. Similar to theNJ2206, theNJ2306 does not show a run--in effect. The full complement CRB, despiteits lower roughness, shows a significant run--in effect. This is not to be traced back to a run--in process in thelip--roller contact but rather to a run--in process in the contact between the rolling elements amongst each

7/28/2019 12FTM02_Soendgen_17122

10/11

10 12FTM02

other. Summarizing theaxial load dependent friction torqueof allconsidered bearing designs areon thesamelevel if they are completely run--in.

Approximation equations

In the scope of this study approximation equations are developed which allow the calculation of the axial load

dependent friction torque of CRB. Only common sense dimensions are used for these equations, the microgeometry of the bearings is neglected. Figure 9 shows the comparison of the results from the approximationequations and the ones from the simulation model according to [9].

The left part of Figure 9 shows the results from the approximation equations whereas the results form thesimulation model are illustrated in the right part. The calculations are based on a NJ2206 bearing which isoperated in the experimentally investigated parameter range. Figure 9 reveals a good correlation betweenthe simulations from the simulation model with the ones from the approximation equations. Only in the rangeof high loads at low speeds the approximation equations lead to higher values than calculated by thesimulation model.

Summary

The aim of this study is the examination of the axial load carrying capacity and of the axial load dependentfriction torque of cylindrical roller bearings (CRB). For this, extensive test runs are performed with five differ-ent bearing designs and three different sizes. The decisive measure is the bearings friction torque. Besidessize and design of the bearings the operation parameters speed, loading and lubricant are varied. The testscomprehend measurements in equilibrium, run--ups, running in and wear experiments. A simulation modelwhich has been developed at the institute allows the calculation of the axial load dependent friction torque,load distribution, Hertzian stresses and lubrication film thicknesses [4] [9]. The results achieved within thisstudy allow the validation of this simulation model. This treatise comprises thedevelopment of approximationequations, which allow the calculation of the axial load dependent friction torque with parameters that areknown to the designer. The results are the basis for the determination of the permissible axial load with a heat

balance of the bearing.

Figure 9. Results from the approximation equations vs. results from the simulation modelaccording to [9]

7/28/2019 12FTM02_Soendgen_17122

11/11

11 12FTM02

References

[1] FAG, Wlzlager auf den Wegen des technischen Fortschritts, Verlag R. Oldenbourg GmbH, Mnchen,1984.

[2] FAG, Hauptkatalog WL41 520/3 DB, 1999.

[3] Bartels, T., Bordreibung von Zylinderrollenlagern Literaturrecherche, Forschungsthema T609,

Forschungsvereinigung Antriebstechnik e.V., Heft 420, 1994.[4] Lubenow, K., Axialtragfhigkeit und Bordreibung von Zylinderrollenlagern, Dissertation und

Schriftenreihe des Instituts fr Konstruktionstechnik, Heft 02.2, Ruhr--Universitt Bochum, 2002.

[5] Schaeffler KG, Wlzlager HR1, 2006.

[6] SKF, Hauptkatalog 5000 G, 2004.

[7] Brndlein, J. Eschmann, P., Hasbargen, L. und Weigand, K., Die Wlzlagerpraxis, Handbuch fr dieBerechnung und Gestaltung von Lagerungen, 3. Auflage, Vereinte Fachverlage GmbH, Mainz, 2002.

[8] Koryciak, J., Einfluss derlmenge aufdas Reibmoment von Wlzlagern mit Linienberhrung, Disserta-tion und Schriftenreihe des Instituts fr Konstruktionstechnik, Heft 07.1, Ruhr--Universitt Bochum,2007.

[9] Koch, O., Dreidimensionale Simulation von kombiniert belasteten Zylinderrollenlagern, Dissertationund Schriftenreihe des Instituts fr Konstruktionstechnik, Heft 08.1, Ruhr--Universitt Bochum, 2008.

[10] Sndgen, S., Verlustleistung und Tragfhigkeit belasteter Borde von Zylinderrollenlagern, Dissertationund Schriftenreihe des Instituts fr Konstruktionstechnik, Heft 09.5, Ruhr--Universitt Bochum, 2009