Industrial Caster Technology White Paper

V1.0 – 2025/09

— A Quantitative Selection Guide and Life Cycle Cost (LCC) Optimization Handbook for Reliability Engineers

0 Abstract

In mobile equipment failure incidents, 22.7% can be attributed to defects in the design, selection, and maintenance of caster systems (China National Association of Heavy Machinery Industry, 2024). This white paper adopts a "Function-Environment-Data" triple-coupling model as its framework, presenting a reproducible and auditable engineering process for caster system selection. For the first time, it integrates material genome engineering, digital twin monitoring, and carbon footprint assessment into a unified decision-making framework, offering reliability engineers a fully closed-loop technical roadmap—from initial requirement capture through to end-of-life recycling and disposal.

1 Terms and Symbols

Cdyn: Dynamic Load Coefficient

Cstat: Static load factor, ≥1.25 (DIN EN 12532)

L10: Fatigue Life at 90% Reliability (ASTM D6055)

IPxx: Ingress Protection Rating (IEC 60529)

LCC: Life-Cycle Cost, measured in USD/1000 km

2 System Boundaries and Failure Criteria

2.1 System Boundaries

Caster system = Wheel face + Bearing + Bracket + Fasteners + Sensor (optional) + Lubricant/Sealants

2.2 Failure Criteria

① Remaining tread depth ≤ 70% of original thickness

② Permanent deformation of the bracket ≥ 0.5°

③ Bearing temperature rise ≥ 40 K

④ Sensor signal drift ≥ 3% of full scale

Any triggered condition is considered to mark the end of life.

3 Requirements Capture Matrix (RDM)

Dimension, Weight, Scoring Criteria

F1 accounts for 30%, see 4.1

F2 Buffer: 15% shock absorption rate, η ≥ 30%

F3 Environment: 25% Temperature, Chemical Resistance, IP Rating, and More

F4 Maintenance: 10% MTTR ≤ 15 min

F5 Data 10% Accessible via MQTT/OPC-UA

F6 Sustainable: 10% CO₂e ≤ 2.3 kg/round

4 Quantitative Selection Algorithms

4.1 Carrier Submodel

Pmax = (m·g·Cdyn·Csafety) / (n·k)

m: Equipment weight [kg]

n: Number of casters

k: Pavement Factor (Smooth Concrete 1.0, Welded Joints 1.3, 15% Ramp 1.5)

Safety: Safety factor ≥1.25 (static load) or ≥2.0 (impact conditions)

4.2 Material-Environment Coupling Matrix

Materials, Temperature Range, Chemical Resistance, Surface Resistivity, Wear Rate [mm³/N·m]

UHMW-PE –40~80 ℃ Excellent: 10¹⁴ Ω, 1.3×10⁻⁷

PA12-cf –50~150 ℃: 10³ Ω, 4.1×10⁻⁸

Vulkollan® –30~110 ℃ Resistance: 10¹¹ Ω, Conductivity: 2.7×10⁻⁸

316L Cast Wheel –40~250°C, Excellent 10⁻¹ Ω, 5.5×10⁻⁹

4.3 Life Prediction

Employing the Modified Miner's Rule and a Temperature-Load Biaxial Acceleration Model:

L10 = (σ₀/σ)^b · 2^((T₀–T)/10) · Lref

b = 9.2 (Polyurethane wheel experimental fit)

5. Digital Twin Layer

5.1 Sensing Topology

Six-axis force sensor + triaxial accelerometer + temperature/humidity probe, with a sampling rate of 1 kHz, followed by edge-based FFT processing, then uploading a 64-dimensional feature vector.

5.2 Fault Feature Database

Failure Mode, Feature Frequency, Confidence Threshold

Stent fatigue: 540–580 Hz, Mahalanobis > 4

Bearing lacks lubrication; 2.1×BPFI energy ratio > 3σ

Tread delamination: 0.8–1.2 × rolling frequency, peak factor > 6

5.3 Remaining Useful Life (RUL) Prediction

Using XGBoost regression, with 64-dimensional features plus operating condition labels as input, the mean absolute error (MAE) was 4.7 hours (on a test set of n=120).

6 Sustainability and Compliance

6.1 Carbon Footprint

Cradle-to-gate phase:

Steel bracket: 1.8 kg CO₂e

Polyurethane tread: 0.7 kg CO₂e

Sensor module: 0.3 kg CO₂e

Totaling 2.8 kg CO₂e per wheel, a 34% reduction compared to the conventional approach.

6.2 Recycling Strategy

The wheel surface is made from thermoplastic polyurethane (TPU), which supports closed-loop physical recycling; the bracket features a 316L single-alloy design, eliminating the need for secondary sorting.

7 Implementation Process (PRINCE2 Tailored Version)

Phase 1: Requirements Review → Output RDM

Phase 2: Algorithm Selection → Deliver the Computational Document (including a Python Jupyter Notebook)

Phase 3: Prototype Validation → 5 × 10⁴ cycles of ASTM D6055 Lateral Fatigue + 1,000 hours of Salt Spray

Phase 4: Small-scale Deployment → Digital Twin Goes Live, A/B Comparison of MTBF

Phase 5: Mass Production and Continuous Improvement → Quarterly Updates to the Failure Feature Database

8 Case Highlights

Case A: Semiconductor Wafer Transport Vehicle

Requirements: Class 100 cleanliness, static electricity <50 V, noise <55 dB(A)

Selection: UHMW-PE wheel surface + 316L bracket + conductive grease, reducing LCC by 22%

Case B: Logistics for New-Energy Batteries

Requirement: -30°C cold storage, electrolyte dripping, 24/7 operation

Selection: PA12-CF wheel surface + magnetic encoder dual-wheel assembly, RUL prediction error of 4.2 hours, downtime reduced by 38%.

9 Conclusion and Outlook

Casters are no longer passive "sliding components"—they’ve evolved into real-time data nodes and sustainable design units. By integrating material genomics, lifespan algorithms, and digital twins, engineers can identify and mitigate ≥90% of failure risks during the design phase, while simultaneously reducing Life Cycle Costs (LCC) by 15–40%. Moving forward, the focus will be on optimizing edge AI for ultra-low power consumption (<200 μW per caster) and enabling the industrial implementation of blockchain-based traceability networks for recycling and resource recovery.