Sei sulla pagina 1di 10

Mechanism and Machine Theory 45 (2010) 14241433

Contents lists available at ScienceDirect

Mechanism and Machine Theory


j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / m e c h m t

Experimental identication of the contact parameters between a V-ribbed belt and a pulley
Gregor epon a, Lionel Manin b, Miha Boltear a,
a b

University of Ljubljana, Faculty of Mechanical Engineering, Akereva 6, 1000 Ljubljana, SI, Slovenia Universit de Lyon, CNRS, INSA de Lyon, LaMCoS UMR 5259, Villeurbanne F-69621, France

a r t i c l e

i n f o

a b s t r a c t
The aim of this paper is to identify the contact parameters between a belt and a pulley that can be used in a two-dimensional multibody belt-drive model. Two experimental setups are proposed in order to identify the contact stiffness and the friction coefcient between the V-ribbed belt and the pulley. The friction coefcient is identied at various initial belt tensions and relative velocities between the belt and the pulley. The measurement procedure and the contact formulation are veried with a numerical experiment. 2010 Elsevier Ltd. All rights reserved.

Article history: Received 1 June 2009 Received in revised form 3 May 2010 Accepted 8 May 2010 Available online 30 June 2010 Keywords: V-ribbed belt Friction Contact stiffness ANCF Multibody dynamics

1. Introduction V-ribbed belt-drive systems have become increasingly important to the automotive industry since their introduction in the late 1970s. Usually, V-ribbed belts in automotive engines drive multiple-accessory pulleys, leading to compactness, smaller pulley diameters and a longer belt life. To ensure stable working conditions the dynamic responses of such systems have been studied extensively. A review of the literature [1] identies two well-dened groups of studies. The rst group deals with the transverse belt span response [2,3] and the rotational response [4] of the pulleys in the belt-drive. The second group deals with describing the belt-pulley contact formulation. Most of the contact models are based on classical creep theory [5] or the shear theory [6]. However, the two groups suffer an unsatisfactory connection with each other: the rst does not take into account the belt-pulley contact behavior and the second neglects the vibration due to the transmission. Leamy and Wasfy [7,8] attemped to bridge this gap between the above-mentioned groups of studies, developing a general, dynamic nite-element model of a belt-drive system, including a detailed frictional contact. This nite-element model was able to predict the belt creep over the pulleys and the belt-drive vibrations. The contact between the belt and the pulley was modeled using a well-known penalty method, together with a Coulomb-like tri-linear creep-rate-dependent friction law. Using the absolute nodal coordinate formulation (ANCF), originally proposed by Shabana [9], the authors in [10] developed a more general planar model of the belt-drive. Most recently, epon and Boltear [11] presented a belt-drive model using the ANCF with a detailed contact formulation between the belt and the pulley. The belt-pulley contact was formulated as a linear complementarity problem (LCP), using the discontinuous Coulomb friction law to model the frictional forces. All the above-mentioned studies [7,8,10,11] describe belt-drive numerical models; however, no work regarding the identication of the belt material and the contact parameters has been presented. For any reliable simulation of a belt-drive the material and the contact parameters should be obtained from experiments. In Ref. [12], epon, Manin and Boltear presented
Corresponding author. Tel.: + 386 1 4771 608 (direct line); fax: + 386 1 2518 567. E-mail address: miha.boltezar@fs.uni-lj.si (M. Boltear). 0094-114X/$ see front matter 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.mechmachtheory.2010.05.006

G. epon et al. / Mechanism and Machine Theory 45 (2010) 14241433

1425

methods for identifying the stiffness and damping of V-ribbed belts. Some researchers have experimentally studied the contact between a grooved pulley and V or V-ribbed belts [13,14]. The traction capacity of V-ribbed belts depends greatly on the distribution of the contact pressure over the ank regions of the belt and the roughness of the contacting surfaces. It was experimentally demonstrated in [13] that with an increasing belt tension the contact area between the belt and the pulley increases. For an accurate description of the contact between a V-ribbed belt and a pulley it is necessary to build a threedimensional nite-element model [14]. However, these models are not appropriate for modeling the belt-drive dynamics over longer time scales as they are computationally inefcient. The aim of this paper is to present methods for the experimental identication of the contact parameters between the pulley and V-ribbed belts. The identied contact parameters will be suitable for incorporation into the planar multibody, belt-drive models presented in Refs. [7,8,10,11]. The procedure includes an experimental measurement of the contact-penalty parameters as well as a measurement of the friction coefcient. As planar belt-drive models are not able to model the belt-rib and pulley-groove geometries, the measured contact stiffness and friction coefcient will include the cumulative contribution of the belt-rubber material as well as the contact geometry. The identied friction characteristics will be used to validate the contact model presented in Ref. [11]. 2. Experimental identication of the contact parameters In this section we will present a method for identifying the contact stiffness and the friction coefcient between a pulley and a V-ribbed belt. The contact parameters will be identied for a V-ribbed belt with ve ribs and a K-rib section (5PK), as shown in Fig. 1. 2.1. Identication of the contact stiffness The penalty-contact model is used in Refs. [7,8,10,11] to compute the normal contact forces between the belt and the pulley. It is implied in Ref. [11], that due to the mainly resting contact between the belt and the pulley (except when the belt enters the pulley), the value of the normal contact forces is almost independent of the value of the stiffness coefcient. However, it does have an inuence on the amount of deformation between the belt and the pulley. It is shown in Ref. [6] that the radial deformation of the belt in the pulley grooves inuences the angular-speed loss between the belt and the pulley. Thus, an accurate identication of the contact stiffness is necessary in order to precisely predict the angular-speed loss between the driver and the driven pulley. In general, the penal normal contact force includes both elastic and dissipative components. The elastic component Fel is a function of the penetration, which in general represents the deformation of the bodies in contact. The dissipative component Fdiss is usually a function of the relative normal contact velocity. The penal force in the normal contact direction can thus be written as: FN = Fel gN + H g N Fdiss g N ; where gN is the amount of penetration, N is the relative normal velocity in the contact and H g N is a step function:  Hx = 1; x 0 : 0; x N 0 2 1

When the belt and pulley come into contact the normal relative velocity is negative N 0 and the pulley penetrates a certain distance gN 0 into the belt. When the contact between the belt and the pulley is established, the belt-rib deforms, which results in an increased contact area between the belt and the pulley, see Fig. 2. As reported in [13], the contact area also depends on the surface roughness in correlation with the normal contact force. Thus, it is expected that the contact stiffness will be a function of the rubber-belt material, the belt-rib and pulley-groove geometries as well as the roughness of the contacting surfaces. In order to measure the cumulative inuence of these parameters on the contact stiffness the experimental setup in Fig. 3 is proposed. The 5PK belt segment of length 4 cm is pressed against the plate, which has identical grooves to the pulley for the K-section belts. The compression of the belt segment and the grooved plate was achieved with the universal Zwick/Roell Z050 testing machine in the temperature chamber. The tests were conducted at four different temperatures, which enabled us to determine the inuence of the temperature on the mechanical properties of the belt-

Fig. 1. Belt segment (5PK).

1426

G. epon et al. / Mechanism and Machine Theory 45 (2010) 14241433

Fig. 2. Contact between the belt rib and the pulley-groove. (

) low normal contact force; (

) high normal contact force.

rubber. In Fig. 4 the compression force versus the deformation of the belt segment is presented. It can be assumed that the main deformation of the belt segment is due to the deformation of the soft friction-rubber layer. Temperatures up to 80 C have practically no inuence on the mechanical properties of the rubber layer, and hysteresis can be observed during the loading and unloading phases. The measurements indicate a nonlinear relation between the force and the deformation. If the curve at 22 C is taken as a reference, the loading and unloading phases can be approximated with a quadratic polynomial, see Fig. 5. The presented curve is normalized to 1 cm of the belt segment length. The elastic component of the penal normal contact force can be written as: Fel gN = 3025:6 10 Lcont gN + 6:5 10 Lcont gN N ;
6 2 5

where Lcont[cm] is the length of the belt segment in centimeters. The value of penetration variable gN is given in [m] and the relative normal velocity is given in [m/s]. Eq. (3) enables us to compute the normal contact forces between the belt and the pulley, depending on the deformation (penetration) between the belt and the pulley. The dissipative component Fdiss of the penalty force is usually a function of the relative normal velocity and is computed with the equation: Fdiss = CP g N; 4

where CP is the damping coefcient. Due to the mainly resting contact between belt and pulley (N 0) the inuence of damping coefcient on normal contact forces is usually negligible [11]. In order to avoid numerical problems and unsurprised vibrations the value of the damping coefcient is chosen to be CP = 300 Ns/m.

Fig. 3. Contact stiffness measurement.

Fig. 4. Deformation of 4-cm-long belt segment versus the compression force. (

) T = 22 C, () T = 40 C, (

) T = 60 C, (

) T = 80 C.

G. epon et al. / Mechanism and Machine Theory 45 (2010) 14241433

1427

Fig. 5. Approximation of the measured contact stiffness at temperature T = 22 C. ( length, ( ) approximation of the contact stiffness with a quadratic function.

) experimentally obtained contact stiffness normalized to 1 cm of belt

2.2. Identication of the friction coefcient The torque transmission between the pulley and the at or V-ribbed belts is mainly governed by the friction. By assuming that the belt is sliding against the pulley and that the friction is fully developed along the groove, we can write the EulerEytelwein equation for the tension ratio: Td =e : Tj 5

The variables Td and Tj are the tensions in the tight and slack belt spans and represents the contact arc between the belt and the pulley. From the tensions in both belt spans the friction coefcient can be obtained as: = ! 1 T ln d ; Tj 6

where is the global friction coefcient, including the inuence of the belt material as well as the belt and pulley-rib proles. The test setup consists of a specially developed test rig, as shown in Fig. 6. In order to measure the tension in the belt spans two load transducers were used. A total of three arcs of contact were studied (1 = 40, 2 = 68, 3 = 82), see Fig. 7. For each contact arc, the tests were performed with eight different initial tensions T0 and ve pulley rotational velocities . The range of the pulley's rotational velocities was deduced from the experiment shown in Fig. 8. The angular-speed loss was measured between the driver

Fig. 6. Experimental setup for measuring the friction coefcient.

1428

G. epon et al. / Mechanism and Machine Theory 45 (2010) 14241433

Fig. 7. Schematic view of the experimental setup for measuring the friction coefcient.

and the driven pulley. From the measured angular-speed loss the maximum sliding velocity between the belt and the pulley can be obtained using the equation: g T;max = loss R; 7

where R is the pulley radius (in our experimental setup R = 0.05 m) and loss is the measured angular-speed loss between the driven and the driver pulley. The measurement of the maximum relative tangential velocity between the belt and the pulley for three different operational conditions is presented in Fig. 9. In order to avoid the inuence of temperature on the friction coefcient the belt segment was tempered to the given temperature before each measurement. A typical force measurement in the tight and slack span is shown in Fig. 10(a) ( = 82, T0 = 270 N). The corresponding friction coefcient, computed using Eq. (6), is shown in Fig. 10(b). It is evident that the coefcient of the sticking friction is slightly higher than the coefcient of the sliding friction. However, as this difference is not signicant the friction coefcient during sliding will be used as the reference friction coefcient ref = sliding. If the reference friction coefcient is constant the normal contact force per unit length can be obtained using the equation [5]: fN = 1 T e ref ; R d 0: 8

In Fig. 11 the analytically computed normal contact force, using Eq. (8), is presented. The mean value of the contact force is assigned to the measured reference friction coefcient. Using this procedure we can identify the inuence of the relative tangential velocity and the normal contact force on the friction coefcient. The values of the friction coefcients obtained for three different contact arcs between the belt and the pulley are presented in Fig. 12. The friction characteristics differ slightly from the value of the contact arc between the belt and the pulley. However, in all three cases the shape of the friction surfaces is almost identical. Clearly, the friction coefcient depends on the normal contact force and the relative tangential velocity, if the dependence of the friction

Fig. 8. Experimental setup for measurement of an angular-speed loss.

G. epon et al. / Mechanism and Machine Theory 45 (2010) 14241433

1429

Fig. 9. Maximum relative tangential velocity between the belt and the pulley versus the torque on the driven pulley: ( ) rotational velocity = 22 rad/s, T0 = 310 N, ( ) rotational velocity = 62.8 rad/s, T0 = 310 N, ( ) rotational velocity = 76.8 rad/s, T0 = 411 N.

coefcient on the normal contact force was expected, the observed strong inuence of the relative tangential velocity on the friction coefcient was not. In Fig. 13(a) the averaged values of all three friction surfaces are presented. The values between the measured points can be interpolated using bicubic splines. The friction characteristics can also be presented in an XY plane (see Fig. 13(b)), where the dependence of the friction coefcient on the normal contact force and the relative tangential velocity is more clearly seen. 3. Numerical belt-drive model The numerical belt-drive model is based on multibody system dynamics together with the absolute nodal coordinate system ANCF [11]. A viscoelastic belt material that obeys the KelvinVoigt model is described by the following constitution relation: l x; t = El x; t + c l x; t ; t 9

where l is the longitudinal distributed stress, E is the Young's modulus, c is the viscoelastic damping factor and l is the Lagrangian strain. The system of the equations of motion, including all the beam elements in the belt and the constraint equations describing the connectivity constraints, can be written as: " MB CeB CeB 0
T

#

e B

::

 =

 Q f + Q eB ; Q dB

10

Fig. 10. Measured belt span axial forces and computed friction coefcient at = 82, T0 = 270 N and = 0.21 rad/s: (a) Axial force in belt spans; (b) Computed friction coefcient with Eq. (6). () axial force in tight belt span, () axial force in slack belt span.

1430

G. epon et al. / Mechanism and Machine Theory 45 (2010) 14241433

Fig. 11. Normal contact force between belt and pulley.

where MB is the constant mass matrix of the belt [9], CeB is the Jacobian of the constraint equations, B is the vector of the Lagrange multipliers. In a belt-drive, the belt is constrained to move over the surface of the pulley. Both the normal reaction force FN and the tangential friction force FT are generated when the belt element contacts the pulley's surface. Each belt element has ve possible contact points, which are equally spaced along the length of the element, see Fig. 14. The normal reaction forces of each contact point are computed using the penalty method, Eq. (1). In the tangential contact direction the Coulomb's friction law on the acceleration level is used: sticking : sliding :
i i jFT jb0 FN i i j = 0 FN jFT

::i gT = 0 ; ::i gT N 0

iIH ;

11

::i where 0 is the coefcient of friction, the set IH contains all possible sticking contacts and g T is the relative acceleration in tangential direction. In order to compute the possible sticking forces the contact problem in the tangential direction has to be formulated as a linear complementarity problem. The equation of motion, including the contact forces between the belt and the pulley, can be written as: qr = HF WN N + WT T + h;
::

12

where N and T are the contact forces in the normal and tangential directions. For a detailed description of the belt-drive model and the contact formulation between the belt and the pulley the interested reader is referred to [11].

Fig. 12. Measured values of the friction coefcient versus the normal contact force and the relative tangential velocity. ( contact arc = 68, ( ) contact arc = 40.

) contact arc = 82, ( o )

G. epon et al. / Mechanism and Machine Theory 45 (2010) 14241433

1431

Fig. 13. Interpolation of the friction surface with bicubic splines: (a) Presentation in space; (b) Presentation in xy plane. ( measured friction coefcients, ( ) interpolation of friction characteristics with bicubic splines.

) averaged experimentally

4. Incorporation of the measured friction characteristic into the numerical model The measured contact parameters can be included into the exible multibody belt-drive model presented in [11]. For each time step of the numerical simulation, depending on the value of the normal contact force and the relative tangential velocity, the friction coefcient is deduced from Fig. 13(a). If the contact parameters (the relative velocity and the normal contact force) are not within the measured friction characteristics the nearest measured value is taken. The numerical modeling of the experiment (Fig. 15) is performed in order to validate the belt-drive model and the contact formulation presented in Ref. [11]. However, with the numerical model of the experiment we simulate the actual experiment for measuring the friction coefcient. In the contact between the belt and the pulley the measured friction characteristics, presented in Fig. 13(a), are applied. The numerical experiment is simulated at the same initial belt tensions and angular velocities of the pulley as in the real experiment. During the simulation, the forces in supports A and B are computed. Finally, using Eq. (6) the friction coefcient, based on the numerical simulation, can be obtained. This friction coefcient can be compared with the one obtained using the actual experiment. The agreement between both friction coefcients can indicate the validity of the numerical model and the friction-measuring procedure. The belt model is based on two-dimensional beam elements, and ten beam elements are used to model the belt segment. In the contact region between the belt and the pulley a ner discretization is applied in order to precisely compute the contact forces. During the numerical simulation for each contact point between the belt segment and the pulley the normal contact force and the relative tangential velocity are computed. Using these data the friction coefcient from Fig. 13(a) can be deduced. The friction coefcients obtained using the numerical simulations are presented in Fig. 16. The values obtained for the different contact arcs between the pulley and the belt are practically the same. In Fig. 17 a comparison of the averaged friction coefcients, obtained with the actual and numerical models of the experiment, is presented. Some differences, especially at low and high normal contact forces, can be observed. These differences between the experiment and the numerical model are due to the simplied expression for the computation of the reference normal force. The

Fig. 14. Belt-pulley contact formulation.

1432

G. epon et al. / Mechanism and Machine Theory 45 (2010) 14241433

Fig. 15. Numerical experiment for validation of the planar contact model.

Fig. 16. Computed values of the friction coefcient with the numerical model versus the normal contact force and the relative tangential velocity. ( contact arc = 82, ( o ) contact arc = 68, ( ) contact arc = 40.

used analytical expression from Eq. (8) neglects the effect of the bending stiffness. This bending stiffness causes the appearance of peaks in the normal forces at the entrance and exit sections of the belt-pulley contact. This means that the reference normal force that is appointed to the measured friction coefcient is not computed with total accuracy.

Fig. 17. Obtained friction coefcient with numerical and actual experiment. (

) actual experiment, ( o ) numerical simulation of experiment.

G. epon et al. / Mechanism and Machine Theory 45 (2010) 14241433

1433

However, we can conclude that with the numerical model and the measured friction characteristics we are able to predict the frictional forces to a sufcient degree of accuracy. Here it is necessary to emphasize that any accurate prediction of the frictional forces relies strongly on the experimentally obtained friction characteristic. From our experience the value of the contact stiffness is not directly proportional to the number of belt-ribs. For instance, from the experimentally obtained contact stiffness for the Ksection belt with ve ribs, it is not possible to accurately predict the values of the contact stiffness for belts that have more or less ribs. Moreover, the values of the contact parameters of the same belt type may also depend on the manufacturer. So, for precise simulations of the belt-drive dynamics using two-dimensional belt-drive models [7,8,10,11], the presented identication of the contact parameters in this paper should be performed. 5. Conclusion In this paper, methods for identifying the contact parameters between a V-ribbed belt and a pulley are presented. Two experimental setups were proposed in order to determine the contact stiffness and the friction coefcient between the belt and the pulley. A nonlinear relation between the normal contact force and the deformation was observed. A quadratic approximation function was used in order to relate the normal contact force with the deformation. The measured friction coefcient depended upon the normal contact forces and the relative tangential velocities. The friction characteristics were interpolated with bicubic splines, which enabled the incorporation of the measured friction characteristics into the numerical model. Finally, the twodimensional belt model and the measured friction procedure were veried with a numerical model of the experiment. Good agreement between the friction characteristics obtained from the actual and numerical experiments was obtained. Acknowledgments This work was sponsored by the Slovenian Research Agency, under contract 3311-04-831674, and the French Rhone-Alpes Region. References
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] S. Abrate, Vibrations of belts and belt drives, Mechanism and Machine Theory 27 (1992) 645659. F. Pellicano, G. Catellani, A. Fregolent, Parametric instability of belts: theory and experiments, Computers & Structures 82 (2004) 8191. G. epon, M. Boltear, Computing the dynamic response of an axially moving continuum, Journal of Sound and Vibration 300 (2007) 316329. S.J. Hwang, N.C. Perkins, A.G. Ulsoy, R.J. Meckstroth, Rotational response and slip prediction of serpentine belt drive systems, Journal of Vibration and Acoustics 116 (1994) 7178. S.E. Bechtel, S. Vohra, K.I. Jacob, C.D. Carlson, The stretching and slipping of belts and bers on pulleys, Journal of Applied Mechanics 67 (2000) 197206. G. Gerbert, Belt slip-a unied approach, Journal of Mechanical Design 118 (1996) 432438. M.J. Leamy, T.M. Wasfy, Transient and steady-state dynamic nite element modeling of belt-drives, Journal of Dynamic Systems, Measurement, and Control 124 (2002) 575581. T.M. Wasfy, M.J. Leamy, Effect of bending stiffness on the dynamic and steady-state responses of belt-drives, ASME 2002 Design Engineering Technical Conferences and Computers and Information in Engineering Conference, 2002. A.A. Shabana, Dynamics of Multibody SystemsThird edition, Cambridge University Press, 2005. K.S. Kerkknen, D.G. Vallejo, A.M. Mikkola, Modeling of belt-drives using a large deformation nite element formulation, Nonlinear Dynamics 43 (2006) 239256. G. epon, M. Boltear, Dynamics of a belt-drive system using a linear complementarity problem for the belt-pulley contact description, Journal of Sound and Vibration 319 (2009) 10191035. G. epon, L. Manin, M. Boltear, Introduction of damping into the exible multibody belt-drive model: a numerical and experimental investigation, Journal of Sound and Vibration 324 (2009) 283296. T.H.C. Childs, D. Cowburn, Contact observations on and friction of rubber drive belting, Wear 100 (1984) 5976. D. Yu, T.H.C. Childs, K.D. Dalgarno, V-ribbed belt design, war and traction capacity, Proceedings of the Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering 212 (1998) 333344.

Potrebbero piacerti anche