CONVEYOR SIMULATION
Single-run dynamic simulation of conveyor belts
By Alex Harrison*, PhD, FIEAust, Conveyor Technologies Ltd., Denver
Velocities and belt tensions are simulated for starting, running and stopping in a single computer run. Inclusion of static and inertial conditions for each mass of a distributed mass-elasticity mechanical model is used to assess possible design problems. Effects of profile, rolling resistance, drive inertia, take-up friction and dispersive wave action can be examined in advance of a final design.
onveyor belts are conducive to analysis using distributed mass-elasticity models. Numerous mechanically equivalent methods have been described in the literature over many years. In the work described here, each mass of a multi-element system is acted on by an acceleration that includes inertia, forces from adjacent spring-coupled masses and body forces due to gravity inclines, take-ups and rolling resistance. Once the acceleration of each mass is defined, element velocities and displacements can be accurately computed. The procedure uses a modified Euler-simulation technique called Euler-X developed to produce very rapidly convergent solutions [1], giving a complete start and stop simulation within a time of between 0.1 s and 1 s. A computer simulation program simultaneously constructs transient forces in the presence of body forces. Simulation of the starting, running and stopping mechanics of a conveyor belt must inherently include both the static (running) and dynamic components of force, velocity and displacement. Alternatives such as wave equation solutions require a superposition of static and dynamic forces as separate computational results. The simulation method used in this paper does not require use of any superposition principles. For general application, the simulation method has been developed as a normalized unit-standard. Each mass, body force and damping component have appropriate ratios relative to a unit belt stiffness. In this way, the model is infinitely scalable. Applications have included an investigation of the influence of rolling resistance on dynamic stress propagation, the effect of sticky take-ups and take-up payout on dynamic surges in longwall conveyors, and the effect of drive inertia on wave amplitudes.
Figure 1.Simulation of starting, running and stopping in a single simulation. Belt speeds and tensions are shown for 8 location (Traces : Black =T1, Red = T2, Blue = Tail, Green = Carry side near tail).
Low Losses
Medium Losses
High Losses
Figure 2. Influence of rolling indentation loss on wave-front amplitudes on breakaway starting.
40
Australian Bulk Handling Review: July/August 2008
CONVEYOR SIMULATION
Figure 3. Take-up pay-out and adjustment simulation, activated about 4 seconds after the belt reaches full speed.
Typical simulation
Predicted velocity and force histories for a 2 km long overland belt are shown in Figure 1. The belt declines slightly then rises with an overall lift of 7 degrees for about half its length, and conveys 2000 t/h at a speed of 3 m/s. The conveyor uses a ST-2500 steel cord belt with a belt modulus E = 180,000 kN/m. A damping-to-stiffness ratio used for the model is c/k = 0.15 s except at the tail where this value is increased for sag effects. Applying this data to an 8-mass model for the carry and return side, the auto-computed static tensions are taken directly from the complete simulation graph of Figure 1 as follows :
T1 (running) = 570 kN; T1 (stopped) = 480 kN; Rolling Loss = 90 kN (Carry). T2 (running) = 187 kN; T2 (stopped) = 237 kN; Rolling Loss = 40 kN (Return). T1 accelerate = 650 kN; S tart Time = 21.8 s; Stop Time = 6.6 s.
be defined by a DIN friction factor of 0.007. A high indentation loss would equate to a DIN factor exceeding 0.03. Note from Figure 2 that a belt with high rolling losses (which includes indentation, belt and material flexure and idler seal drag) is more prone to some initial runback on breakaway at low-tension areas. Here, the initial velocity wave amplitude increases in intensity, as does the starting tension. In effect, the belt is resisting displacement in the presence of a tension change. An overall low combined rolling resistance will result is a smoother, less dynamic start of the belt.
Winches with friction
Other dynamic effects can occur in conveyors that have active tensioning. For example, winches located behind the drive at T2 often show unpredictable slewing which results from payout of belt into the system as material load-on occurs. If the winch carriage or rope sheaves contain sufficient friction, a winch may not pay out or pull in immediately in response to belt tension changes at T2. The delay in winch response can cause drive slip or momentary belt speed changes on take-up pay-out when friction resists take-up carriage motion [2]. Underground long-wall belts that are similar to the example belt often exhibit erratic take-up position changes. Figure 3 shows a simulation of a winch paying out just after the example belt has reached full speed. In this situation, a dynamic increase in belt speed will be observed until tensions are recovered in the T2 area. The simulation shows the effect on T1 and T2 tension for the example in which the tail area winch activates just after full speed. If a winch were located at T2, any slack belt resulting from take-up pay-out will launch a wave front and may cause a low tension pulse to be propagated to the tail of the belt. Such a condition can also occur when winches and capstan combinations are used to maintain a nominal take-up mass position between limits, particularly if the limit control is affected by friction in the mechanical structure. The takeup motion is not shown on the figure, but is easily displayed from take-up mass displacement arrays when required.
Wave action occurs on starting and stopping. The stopping time is not an input to the model, since the model computes the stopping impulse. On starting, the tail of the conveyor near the load point reverses slightly due to differences between carry and return rolling losses for the initial belt tensions. Elastic wave speed is determined from the tail start time-delay, which is about 0.9 s, giving an average elastic wave-front speed of 2,220 m/s. On the carry side near the tail, the belt tension approaches zero on stopping. In this situation, the design would require an increase in take-up pre-tension. A nominal tail tension was set at 92 kN for the initial run.
Breakaway and rolling losses
An important application of the mechanical simulation approach is an ability to show the influence of rolling resistance on wave amplitude in the velocity domain. Low rolling resistance belt covers result in a less severe dynamic impact on breakaway. Several runs of the model for very low and abnormally high indentation rolling resistance indicates that the breakaway wave front amplitude increases as indentation resistance rises, shown in Figure 2 for the example conveyor. The model allows a breakaway value for idler seals as a separate input to the static running idler drag which is a function of normal loading and speed, whereas the visco-elastic indentation loss is considered to be constant for any given rubber type, since the wave front speed considerably exceeds the indentation rate at each idler. Low indentation loss would
Drive inertia effects
Another factor that influences the dynamic forces and mass displacements, particularly on stopping, is the rotating drive mass inertia translated to the belt line. To study this effect,
Australian Bulk Handling Review: July/August 2008
41
CONVEYOR SIMULATION
Conclusions
Dynamic simulation of conveyor belts requires mechanical analysis of the belt to derive the equations that govern element accelerations. Once developed, the model has broad applications for the simulation of many conditions of real conveyor operating and control conditions. The unit-standard model used in this paper is scalable to any input variable range. The simulations discussed in the paper shed new light on the way wave fronts propagate in the presence of rolling resistance. Much of the research into the dynamic behaviour of belts over many years can be explained using modelling techniques shown in the paper, such as the effects of drive inertia on wave severity during stopping. Stopping dynamics has a substantial impact on winch slew rate and control design. Modelling dynamic behaviour of belts is a challenge that required a model that is based on sound mechanical principals, is guided by field test results wherever possible, and results in a mathematically stable simulation program. The use of simulation to derive control dynamics and to investigate non-linear and friction related effects in real systems has been discussed and is an ongoing area of application.
M drive = 40000 kg
M drive = 80000
M drive = 180000 kg
Figure 4. Effect of drive inertia on dynamic wave action.
the stopping dynamics of the example belt is shown in Figure 4 for 3 equivalent drive masses. New theories have been developed to allow modelling of the effects shown by both Figures 2 and 4, however it is sufficient here to present typical simulation findings. In Figure 4, the drive mass input is a model variable. For stopping simulation, the model derives a forcing function (the impulse) which includes the drive mass. Stopping dynamics at 8 locations along the carry and return belt runs are computed from the model and used to display the different effects of drive inertia. For the case of a low drive inertia, the initial stopping velocities are rapid and the introduction of third-order vibration modes is evident. These effects will place greater demand on, say, a winch take-up to respond or slew on stopping if the winch is active. For the case of high drive inertia, tensions near the tail become more positive and a winch take-up would need to slew less rapidly. Sometimes, a longer stopping time caused by increased drive inertia may exceed the stop time of out-by belts, or local safety laws. In each case, the analysis process works to allow design parameters to be established.
Other references by the author
[1] (2008). Non-Linear Belt Transient Analysis. Bulk Solids Handling. Vol. 23, No.4, pp 240 245 (in press) [2] (1992), Modern Belt Take-ups and Their Dynamic Motion. Bulk Solids Handling, Vol. 12., No.4, pp 581-584 (1992) * Alex Harrison. PhD, FIEAust, Formerly Professor in Mechanical Engineering Dept., MERZ/TUNRA Chair of Bulk Handling and Conveying, University of Newcastle, NSW and currently Manager of Conveyor Technologies Ltd. LLC in the USA.
Contact: aharrison@conveyorscience.com or conveyorscience@aol.com
42
Australian Bulk Handling Review: July/August 2008