2. Belt dynamics fundamentals

Every multi-stage belt control strategy ultimately manipulates one physical quantity: the distribution of tension along the belt path. Speed references, torque limits, and take-up positions are only indirect handles. To design a controller that keeps a multi-stage conveyor stable, you must first understand how tension propagates, how the belt stretches, and why the belt never behaves like a rigid chain.

Tension as the State Variable of a Conveyor

Consider a single drive pulley with wrap angle and a belt running from a tail pulley. Two tensions coexist at the drive pulley: the tight-side tension T1 on the strand leaving the pulley in the direction of motion, and the slack-side tension T2 on the returning strand. The difference drives the belt:

F_drive = T1 - T2
P = F_drive * v = (T1 - T2) * v

The ratio T1/T2 is bounded by the Euler–Eytelwein (capstan) relationship:

T1 / T2 <= e^(mu * theta)

where mu is the friction coefficient between belt and pulley and theta is the wrap angle in radians. For a typical steel-cord belt on a lagged pulley, mu ≈ 0.35 and theta = 3.5 rad gives e^(mu*theta) ≈ 3.4. If the controller commands more torque than this ratio permits, the belt slips — the pulley rotates but the belt does not advance at the same speed.

Along the carrying run, tension is not constant. It changes with belt weight, material load, incline, and idler resistance. For a straight inclined section of length L, inclination beta, belt mass per metre qB, and material mass per metre qM:

T(x) = T_start + (qB + qM) * g * sin(beta) * x + f * (qB + qM) * g * cos(beta) * x

where f is the artificial friction coefficient (typically 0.02–0.03 for well-maintained idlers). This linear growth is why a multi-stage conveyor cannot be treated as a single tension node: each stage has its own local tension profile, and the coupling between stages is what makes control difficult.

Elongation and the Elastic Modulus

Belt elongation under tension follows Hooke's law in the elastic regime:

epsilon = T / (E * A)

where E is the effective elastic modulus of the belt carcass and A is the cross-sectional area. For a fabric belt, E is typically 300–800 N/mm per ply; for a steel-cord belt, E can reach 10,000–20,000 N/mm. This difference matters enormously: a steel-cord belt stretches very little, so a small change in take-up position produces a large tension change. A fabric belt stretches more, which gives the controller more "slack" but also more transient oscillation.

The total elongation of a belt section is the integral of strain along its length. In a multi-stage system, the take-up device must absorb the sum of elastic elongations plus the permanent elongation that accumulates over time. If the take-up stroke is undersized, tension will drift upward as the belt creeps, eventually exceeding the safe working tension.

Creep: The Slow Relative Motion

Creep is the elastic stretch and recovery of the belt as it passes over the drive pulley. On the tight side, the belt is stretched by T1; on the slack side, it contracts to T2. The belt therefore moves slightly faster on the tight side than on the slack side. The creep velocity is:

v_creep = v * (T1 - T2) / (E * A)

For a steel-cord belt with E*A = 2.0e7 N, T1 - T2 = 200 kN, and v = 4 m/s, the creep velocity is about 0.04 m/s — roughly 1% of belt speed. This is normal and unavoidable. It becomes a problem when creep is mistaken for slip, or when the controller tries to eliminate it by over-tensioning the belt.

Slip: When Friction Fails

Slip occurs when the required driving force exceeds the available friction. The belt then slides over the pulley, generating heat, wear, and loss of speed control. In a multi-stage system, slip in one stage changes the tension in adjacent stages, which can cascade into a full-system stall.

Slip is detected by comparing belt speed to pulley surface speed. A speed deviation greater than 2–3% under load is a reliable indicator. The controller should respond by reducing torque demand or increasing take-up tension, not by increasing drive torque, which would worsen the slip.

Viscoelastic Behavior

Rubber and fabric belts are viscoelastic: their stress response depends on strain rate and history. Two consequences dominate multi-stage control:

• Stress relaxation: under constant strain, tension decays over time. A belt held at fixed take-up position will lose tension over hours.
• Hysteresis: the loading and unloading paths differ, so energy is dissipated as heat. This is why belt temperature rises during high-tension operation.

A practical model for tension decay under constant strain is:

T(t) = T0 * exp(-t / tau)

where tau is the relaxation time constant, typically 10–60 minutes for rubber belts. This means a controller that sets tension once and forgets it will drift out of specification. Periodic tension re-establishment or closed-loop take-up control is mandatory.

Why This Matters for Multi-Stage Control

In a single-stage conveyor, tension control is local. In a multi-stage system, each stage's tension affects the next stage's entry tension, and the belt itself couples them through its elasticity. A step change in Stage 1 drive torque propagates as a tension wave through the belt at the wave speed:

c = sqrt(E / rho)

For a steel-cord belt with E = 1.5e10 Pa and rho = 7800 kg/m³, c ≈ 1400 m/s. Over a 500 m stage, the wave transit time is about 0.36 s. This is fast enough that a naive PID controller on Stage 2 will see the disturbance almost immediately, but slow enough that the two stages can oscillate against each other if their control loops are not coordinated.

Practical Implications for Controller Design

Three design rules follow directly from the physics above:

1. Never control tension by torque alone. Torque sets the tension difference T1 - T2, not the absolute tension level. Absolute tension is set by the take-up device. A controller that only adjusts drive torque will drift in absolute tension as the belt relaxes.
2. Coordinate stages through tension, not speed. Speed references are useful for production rate, but the coupling between stages is tension. A Stage 2 controller that reacts to speed error will fight Stage 1. A controller that reacts to entry tension error will cooperate.
3. Account for creep in speed matching. If Stage 1 and Stage 2 must run at the same linear speed, the controller must compensate for the ~1% creep difference between tight and slack sides. Ignoring this causes material accumulation or starvation at the transfer point.

Understanding these fundamentals is not academic. Every tuning decision, every alarm threshold, and every take-up specification in a multi-stage system traces back to the tension–elongation–creep–slip relationship. The next section builds on this foundation to show how these quantities are measured and used in closed-loop control.