Ruffini, Marco (2025) An integrated virtualization framework for tire manufacturing and testing: from curing to performance. [Tesi di dottorato]

	Documento PDF Marco_Ruffini_38_COMPLETO.pdf Visibile a [TBR] Amministratori dell'archivio Download (31MB) | Richiedi una copia
	Documento PDF Marco_Ruffini_38_PARZIALE.pdf Visibile a [TBR] Amministratori dell'archivio Download (5MB) | Richiedi una copia

Abstract

The tire manufacturing industry is experiencing a paradigm shift driven by the convergence of stringent performance requirements, sustainability goals, and accelerated development timelines. Traditional tire design has long relied on extensive experimental testing and prototyping, which, while effective, entails high costs, long lead times, and significant energy consumption. The growing need for predictive, resource-efficient, and environmentally sustainable methodologies has led to the increasing adoption of simulation-based approaches. Within this context, this doctoral dissertation introduces an integrated virtual modeling framework that combines physics-based formulations, data-driven methods, and state-estimation techniques to optimize tire manufacturing and performance across multiple domains. The overarching goal is to reduce reliance on full-scale experimental campaigns while improving the interpretability, accuracy, and industrial relevance of the resulting procedures. The first contribution concerns the development of a physics-based thermal model for the curing process, capable of predicting in real-time, directly during the curing cycle, the thermal evolution within each tire layer, using only standard inputs from the curing press machine and the thermophysical characterization of the tire compounds. This virtual approach provides an alternative to intrusive thermocouple-based measurements, allowing accurate internal temperature simulation while reducing setup complexity, instrumentation costs, and production variability. It enables assessment of cure uniformity, identification of thermal gradients, and optimization of process parameters such as dwell time and mold temperature. Moreover, by linking process parameters to compound behavior, the model facilitates improvements in material formulation and energy-efficient curing strategies. Finally, this model, if coupled with suitable curing kinetic equations, can form the physical foundation of a digital twin of the curing process. The second contribution presents a predictive framework for tire rolling resistance based on a master curve derived from a limited number of controlled drum tests. Starting from a few reference experiments, the methodology establishes a master curve linking rolling-resistance torque to the tire’s internal temperature field. Then, by identifying horizontal and vertical shift laws, the dataset can be virtually expanded to cover a wide range of load, inflation pressure, speed, and temperature conditions, effectively virtualizing most of the traditional drum-testing campaign and reducing the need for physical tests. This enables estimation of rolling resistance not only under standard homologation conditions but also across a broader range of real operating scenarios, supporting faster optimization of tire performance with respect to CO2 emissions and fuel consumption. Moreover, by coupling the master-curve formulation with temperature measurements from onboard sensors or with temperature estimates obtained through an Extended Kalman Filter, the framework can be extended beyond indoor use, enabling real-time evaluation of rolling losses directly from sensor data or estimated thermal states. In this way, the resulting virtual model also provides the foundation for eco-driving applications, offering predictive insight and real-time feedback to foster more energy-efficient driving behaviors. The third contribution focuses on modeling the tire–road contact patch, the small yet fundamental interface at the core of tire–road interaction, whose accurate description is crucial for tire performance modelling, as the footprint geometry defines the boundary conditions for both mechanical interaction and thermal exchange with the road. To this end, a regression-based framework is developed to reproduce and generalize the contact patch geometry across a wide range of operating conditions, including variations in vertical load, inflation pressure, and camber angle. Starting from a limited set of experimentally acquired static contact patches, the footprint shape is described using a superellipse formulation that compactly captures both symmetric and asymmetric shapes with a small number of physically meaningful parameters. These parameters are then correlated with the main operating variables through multivariate and symbolic regression, yielding analytical expressions capable of predicting the footprint geometry under intermediate or untested conditions. This approach effectively virtualizes the contact patch characterization process, reducing experimental effort while preserving physical interpretability and ensuring seamless integration into higher-level tire performance simulations. The final contribution addresses one of the most safety-critical and experimentally demanding aspects of tire behavior: braking on wet roads. The dissertation introduces a systematic data-driven methodology to quantify the influence of material, structural, and geometric factors on wet-braking distance. A comprehensive feature-selection framework is implemented to identify the most influential predictors of wet grip, isolating the respective effects of tread viscoelasticity, tread pattern, and contact-patch geometry while clarifying the physical mechanisms governing friction generation under wet conditions. Building on this, a regression model is developed to provide accurate and interpretable predictions of braking distance from a compact set of tire and material parameters. Unlike conventional black-box machine-learning models, the proposed approach preserves physical interpretability, expressing the relationships between tire features and wet-braking performance through explicit formulations. Once calibrated, the model enables the virtualization of wet-braking tests, allowing the prediction of braking distances and thus drastically reducing the need for physical experiments. In addition, the resulting model enables rapid sensitivity analyses, supports compound and tread optimization, and provides clear insights into how specific viscoelastic parameters affect wet grip. Collectively, these contributions constitute a coherent virtual testing platform that integrates physical modeling, statistical learning, and state estimation within a unified methodology. Each component addresses a key aspect of tire technology, adhering to consistent methodological principles of interpretability and physical rigor. Together, they enable engineers to improve tire manufacturing and development in a fully virtual environment. By reducing reliance on physical prototypes, accelerating development, and enhancing the physical understanding of tire behavior, the proposed framework advances the tire industry toward digitalized, eco-efficient, and predictive design processes aligned with the future demands of sustainable mobility.

Downloads

Actions (login required)

Modifica documento