Experts from the University of Newcastle and Elastotec investigate the emerging issues around pulley friction, the complexity of the issue, and the failures that may result.
Belt conveyors form a critical part of the materials handling process and have established themselves as the chosen technology for transporting bulk materials at high tonnages over long distances.
As the scale of new mining operations continues to increase, this has resulted in an unprecedented demand on these systems to convey material further, faster, and up greater elevation.
Significant improvements in the tensile strength of belts and the refinement of dynamic models to assess the transient tensions under starting and stopping conditions have allowed for reduced safety factors, greater transport distance and lift, as well as faster belt speeds. Despite the considerable amount of research that made these achievements possible, much room for improvement remains in understanding the effectiveness of drive systems to transmit the large amounts of power now required by these systems.
The drive system forms a critical component of any belt conveyor, tasked with transmitting the force required to start and stop the conveyor, as well as maintain a constant operating velocity. This force, known as the effective belt tension, is traditionally transmitted through a drive pulley, to the pulley lagging (if installed), to the rubber bottom cover of the conveyor belt, and eventually to the reinforcing carcass (steel cord or fabric) within the conveyor belt. The effectiveness of this transmission is ultimately defined within a frictional contact between the conveyor belt and the drive pulley surface.
Review of current design methods
Drive system design relies on the respective belt tension either side of the drive pulley, denoting T1 as the tight side tension, and T2 as the slack side tension. The difference between these two represents the effective tension (Te) outlined above, and the force required to be transmitted through the drive system to allow the belt to operate. The design method is well understood, based on Euler’s classic ‘rope friction’ model defined below.
Where, µ = the coefficient of friction between the belt and the pulley surface, and
θ = the angle of wrap of the belt around the pulley
The simplicity of this model naturally has corresponding limitations, the majority of which are also understood within industry. The assumption that the friction is fully developed around the arc of contact relates only to rigid body contact and means that each point of contact around the pulley surface exhibits an equal coefficient of friction. This is not possible for a viscoelastic drive mechanism in shear. The gradual increase in belt tension around the drive pulley, coupled with the varying normal force (see Figure 1) results in a varying degree of viscoelastic slip within the contact, and therefore a varying coefficient of friction. For viscoelastic surfaces, friction may only truly reach is maximum kinetic value when the belt is slipping, or on the verge of slipping. Design standards such as ISO5048 and DIN22101 compensate for this by utilising a conservative coefficient of friction of up to 0.45, depending on lagging type and degree of contamination, while traditional values of kinetic friction may be as high as 0.8 for ceramic lagging .
Actual friction behaviour around the pulley arc
Several researchers have investigated the actual friction behaviour around the pulley arc. Work conducted by Harrison  developed a FEM model of the conveyor belt around the arc of the pulley, to determine the change in tension around the arc, and an effective coefficient of friction based on standard input values of 0.9 and 0.7 for static and kinetic friction respectively. His work concluded that the belt tension around the pulley displayed an s-curve response as shown in Figure 2, based on the ratio
k= T1 /T2
, with the effective coefficient of friction dependent on this ratio.
Given a value of
, Harrison determined an effective coefficient of friction of 0.32 to 0.84, dependent on the tension in the system. A fabric belt typically corresponding to lower tensions yields a lower effective value, compared to a higher tension steel cord installation.
Work by Tiwari et al  investigated the viscoelastic dependency (load and slip rate) on the coefficient of friction of ceramic lagging tiles on a rubber belt. Ceramic lagging is commonly used in applications at risk of contamination, as the lagging has cylindrical protrusions extruded from the surface, which indent the rubber cover to aid traction. Naturally, this indentation is load and slip dependent, with Tiwari predicting a coefficient of friction in a planar contact region (flat surfaces) of around 0.4 – 0.8 depending on rate of slip between the two surfaces. When the ceramic protrusions begin to be contaminated by rubber shavings from belt wear, this coefficient increases to as high as 1.5 under a contact pressure of 0.62 MPa. Coefficients of friction in excess of 1 have also been measured experimentally by Robinson et al  for both ceramic and polyurethane lagging under a curved contact area. It is well known that these coefficients will vary with age and wear of both the pulley lagging and belt cover.
What about across the pulley face?
A more recent development is the need to develop a deep understanding of the behaviour of friction across the pulley face. Euler’s equation, as well as much of the existing research focus on understanding the behaviour around the arc of contact, operating on the assumption that the viscoelastic behaviour, and thus the coefficient of friction is uniform across the pulley width. For a new installation, this assumption is valid, however as belt and lagging wear, the belt thickness and/or pulley diameter becomes less uniform, leading to stress concentrations within the belt being transferred to the pulley lagging. Build up or contamination of lagged surfaces has the same effect.
Conveyor belts wear for a variety of reasons on both sides. The carry side of the belt (top cover) typically wears most towards the centre, at the surface in contact with the bulk material. Similarly, the return side (bottom cover) will typically exhibit localised wear associated with wing idler pressure, idler junction wear, idler skew, or any combination resulting in an uneven cover thickness. When a worn steel cord belt travels around a pulley, the arc length of each steel cord will vary across the belt width, resulting in localised stresses on the cords and splices. This stress is subsequently transmitted to the lagging, and its bonding with the pulley.
The resulting cord stresses are not directly proportional to the length of arc as one may expect. To maintain total load across the belt, cords adjacent to those with reduced load bearing compensate for the reduction in load capacity of the worn sections, leading to a peak ‘switch’ stress at the wear junction. This stress is subsequently transmitted to the pulley lagging, and ultimately the bonding method, potentially leading to failures on lagged drive or high-tension bend pulleys, as well as reduced life expectancy of belt splices.
Research conducted by Overland Conveyor and Elastotec  has developed a model to predict the increase in stress due to unequal load sharing between cords, based on Hedgepeth’s model for cable load sharing . An example from this research is given below, describing the load sharing between cords for an 1800-millimetre wide belt, consisting of 120 cords across its width. A parabolic wear pattern 10-millimetres deep at its centre is removed from the middle third of the belt, to simulate carry-side wear.
As can be seen from this research, the switch stress is located at the wear junction point, and in this instance, is approximately double that of the uniform cord tension.
Implications for pulley design
Drive and high-tension bend pulleys typically experience the highest tension in a conveyor system, and as such small changes in cord path length due to cover wear or pulley contamination leads to significant changes in the load sharing between cords. Some examples are highlighted below.
High tension bend pulley: iron ore
Ceramic lagging on a high-tension bend pulley began to show premature damage along the centreline of the pulley. The installed belt exhibited cover wear along the central section of the carry side, due to wear from the bulk material. This change in belt thickness resulted in an increased stress along the centreline of the high-tension bend pulley, and subsequently failure of the ceramic bonding.
High tension bend pulley: copper
In a copper mine, signs of physical damage and delamination along the centreline of a high-tension bend pulley began to show after three months in operation. The installed belt exhibited cover wear along the central section of the carry side, resulting in increased stresses along the centreline of the high-tension bend pulley. This increase in stress lead to the increase in wear and subsequent delamination of the lagging.
Drive pulley: Gold
Upgrade of a conveyor belt in a gold mine to a higher strength, resulted in a significant increase in lagging wear, despite the same lagging being used successfully on the previous belt, and the same operating conditions of the conveyor. Drive and lagging design are all within current design specifications. While the cause of this failure is still under investigation, it is believed that an increase in stress was transferred to the lagging, resulting from the upgrade of the conveyor belt.
The existence of uneven load sharing between steel cords would likely present subtly in two ways; uneven lagging wear on the pulleys, and/or mis-tracking of the conveyor belt. This would develop with time, and without rupture or damage of individual cords, is unlikely to be detected by current belt monitoring systems. Detection of this sooner rather than later would help minimise fatigue to individual cords and splices and extend the service life of the belt.
The future of research
In response to these cases, Elastotec has partnered with the University of Newcastle to further develop the understanding of friction behaviour within a pulley contact. Experience from both companies has demonstrated that when it comes to friction of viscoelastic bodies, tests utilising static or rigid body dynamics can lead to erroneous outcomes, and in particular, curved surfaces behave differently compared to flat. To completely understand this behaviour, a full-scale test facility is currently being designed, and will be built at the University of Newcastle, with the capabilities of measuring crucial properties around the pulley circumference, and across the pulley face. Parameters such as normal load/pressure, shear force, and belt tension are just a few that will be measured, for a given belt tension, lagging surface and angle of wrap. Tests will be conducted at installed operating tensions. Testing will be completed for a variety of lagging types, on new and used belt samples. In addition to this, used belt samples will be tested on both sides, to measure stresses around a high-tension bend pulley in contact with the carry side.
By extension, once the stress profile within the contact is understood it will then be possible to review existing lagging installations, and determine the maximum stress exhibited in these materials. This knowledge will allow a performance envelope to be defined for each lagging material and validated using the planned test facility. It will also be possible to extrapolate to the limit of allowable power transfer for a given drive design.
The inadequacy of a simple Euler friction model to predict friction behaviour around a pulley has long been known, however the extent of this misunderstanding is only currently coming to light. Beginning as a means of preventing belt slip in a drive pulley, the complications associated with a lack of understanding of the drive traction can and is leading to failures of belt covers and lagging materials.
The intent of this paper is to raise awareness of emerging issues around pulley friction, the complexity of the issue, and the failures that may result. A deeper understanding of friction and the stresses that result will allow engineered design of lagging materials and optimise lagging selection. As conveyors continue to grow in length and throughput, designers will have confidence in the lagging selection, and its performance under new and aged conditions. These topics are the subject of ongoing research.
1 A. Tiwari and B. N. J. Persson, “Conveyor Belt Drive Friction: Ceramic Lego Sliding on Rubber Belt.” PGI-1, FZ Juelich, Germany.
2 A. Harrison, “Modelling Belt Tension Around a Drive Drum,” Bulk Solids Handl., vol. 18, no. 1, pp. 75–79, 1998.
3 P. Robinson, C. Wheeler, V. Agarwal, W. Srinivas, and J. Guo, “Pouch conveyor drive system dynamics,” Int. J. Mech. Mater. Des., vol. 17, no. 1, pp. 187–199, 2021.
4 A. Reicks, “R&D Report – Theoretical Model to Evaluate Belt Wear Effects on Lagging Stress.” Overland Conveyor Company, Elastotec Pty Ltd.
5 J. M. Hedgepeth, Stress concentrations in filamentary structures. National Aeronautics and Space Administration, 1961.