The design and development of high-performance carbon-carbon (C/C) braking systems for motorsports require navigating complex, non-stationary multiphysics phenomena, including thermodynamics, structural mechanics, and tribology. While fully coupled thermoelastic finite element method (FEM) models capture these interactions with high fidelity, their computational cost makes them impractical for simulations of complete braking manoeuvres and full laps during early-stage design iterations. To bridge this gap, this dissertation presents the development and validation of a novel, computationally efficient coupled thermoelastic 3D finite volume method (FVM) model, conducted as a joint collaboration between Politecnico di Milano and Brembo Racing. Implemented in Dymola using the non-causal Modelica language, the framework utilises equivalent material properties derived from solid to void volume and surface ratios to effectively capture the complex geometries of ventilated discs and pads without the computational overhead of fine meshing. A local modelling approach is adopted, incorporating an analytical and semi-empirical friction law inferred from global dynamometer data to predict local torque contributions based on contact pressure, temperature, and sliding speed. The proposed thermoelastic 3D FVM model was validated against full-scale experimental dynamometer data for standard and wedge-amplified Formula One callipers up to 40° wedge angles. Across diverse operating conditions, including mapping, heavy duty, high torque, and telemetry cycles, the model demonstrated high predictive accuracy. Braking torque predictions yielded normalised peak torque and normalised RMSE errors below 5% for most of the configurations and cycles analysed, while disc surface temperature estimations maintained a RMSE below 50 °C in all conditions. The validated model is applied to predict the performance of a new and untested prototype, designed to deliver the same theoretical braking torque at the same hydraulic pressure as an existing reference braking system. The agreement with both the simulated and experimental results of the reference calliper further suggests the capability of the model as a flexible tool to evaluate the braking performance of early-stage designs, based on their geometry, hydraulic layout, and wedge angle. To assess component impact on overall track performance, the research with Brembo Racing was extended to a vehicle-level co-simulation framework coupling VI-CarRealTime with MATLAB/Simulink. This integration overcame native software limitations by enabling dynamic brake balance management and accurate disc temperature tracking over complete race stints, with the majority of temperature percentage errors between 3% and 8%. Recognising the absence of local friction data for racing-specific C/C composites, a custom pin-on-disc tribometer capable of handling contact pressures up to 5 MPa was mechanically designed and assembled during a joint collaboration with Cranfield University. This hardware lays the essential groundwork for future experimental characterisation of C/C tribology, which will ultimately replace the inferred analytical local friction law to possibly further refine the Dymola model’s predictive fidelity.
La progettazione e lo sviluppo di sistemi frenanti ad alte prestazioni in carbonio-carbonio (C/C) per il motorsport richiedono la gestione di complessi fenomeni multifisici non stazionari, che includono la termodinamica, la meccanica strutturale e la tribologia. Sebbene i modelli termoelastici completamente accoppiati basati sul metodo degli elementi finiti (FEM) riescano a catturare queste interazioni con elevata fedeltà, il loro costo computazionale li rende impraticabili per la simulazione di frenate complete e di interi giri di pista durante le fasi iniziali di progettazione. Per colmare questa lacuna, la presente tesi illustra lo sviluppo e la validazione di un nuovo modello termoelastico accoppiato 3D, computazionalmente efficiente e basato sul metodo dei volumi finiti (FVM), realizzato in collaborazione tra il Politecnico di Milano e Brembo Racing. Implementato in Dymola utilizzando il linguaggio acausale Modelica, l’ambiente di simulazione sfrutta proprietà equivalenti dei materiali, derivate dai rapporti vuoto-pieno di volume e superficie, per replicare efficacemente le complesse geometrie dei dischi ventilati e delle pastiglie, senza l’onere computazionale di una mesh fine. Viene adottato un approccio di modellazione locale, che incorpora una legge di attrito analitica e semi-empirica (dedotta dai dati globali del banco prova dinamometrico) al fine di prevedere i contributi di coppia locale in funzione della pressione di contatto, della temperatura e della velocità di strisciamento. Il modello termoelastico 3D FVM proposto è stato validato rispetto a dati sperimentali ottenuti sul banco dinamometrico, relativi a pinze Formula 1 standard e con amplificazione a cuneo (con angoli fino a 40°). Attraverso diverse condizioni operative, inclusi cicli di mappatura, heavy duty, high torque e di telemetria, il modello ha dimostrato un’elevata accuratezza predittiva. Le previsioni della coppia frenante hanno prodotto un errore di picco normalizzato e un errore RMSE normalizzato inferiori al 5% per la maggior parte delle configurazioni e dei cicli analizzati, mentre le stime della temperatura superficiale del disco hanno mantenuto un RMSE inferiore a 50 °C in tutte le condizioni. Il modello validato è stato inoltre applicato per prevedere le prestazioni di un nuovo prototipo non ancora testato, progettato per fornire la stessa coppia frenante teorica alla medesima pressione idraulica di un sistema frenante di riferimento esistente. La concordanza con i risultati sia simulati che sperimentali della pinza di riferimento conferma ulteriormente la validità del modello come strumento flessibile per valutare le prestazioni in frenata di progetti in fase preliminare, sulla base della loro geometria, del layout idraulico e dell’angolo del cuneo. Per valutare l’impatto dei componenti sulle prestazioni complessive in pista, la ricerca con Brembo Racing è stata estesa a un ambiente di co-simulazione a livello di veicolo, accoppiando VI-CarRealTime con MATLAB/Simulink. Questa integrazione ha superato le limitazioni native del software, consentendo una gestione dinamica della ripartizione di frenata (brake balance) e un tracciamento accurato della temperatura dei dischi per interi stint di gara, con la maggioranza degli indici d’errore percentuale sulla temperatura compresi tra il 3% e l’8%. Riconoscendo infine l’assenza di dati sull’attrito locale per i compositi C/C specifici per le competizioni, è stato progettato e assemblato un tribometro pin-on-disc personalizzato, in grado di gestire pressioni di contatto fino a 5 MPa, durante una collaborazione con la Cranfield University. Questo hardware pone le basi essenziali per la futura caratterizzazione sperimentale della tribologia dei compositi C/C, che andrà infine a sostituire la legge analitica di attrito locale dedotta, al fine di perfezionare ulteriormente la fedeltà predittiva del modello in Dymola.
Thermoelastic Finite Volume Method (FVM) modelling and experimental testing of high-performance carbon-carbon (C/C) braking systems
GALVANINI, GIACOMO
2026
Abstract
The design and development of high-performance carbon-carbon (C/C) braking systems for motorsports require navigating complex, non-stationary multiphysics phenomena, including thermodynamics, structural mechanics, and tribology. While fully coupled thermoelastic finite element method (FEM) models capture these interactions with high fidelity, their computational cost makes them impractical for simulations of complete braking manoeuvres and full laps during early-stage design iterations. To bridge this gap, this dissertation presents the development and validation of a novel, computationally efficient coupled thermoelastic 3D finite volume method (FVM) model, conducted as a joint collaboration between Politecnico di Milano and Brembo Racing. Implemented in Dymola using the non-causal Modelica language, the framework utilises equivalent material properties derived from solid to void volume and surface ratios to effectively capture the complex geometries of ventilated discs and pads without the computational overhead of fine meshing. A local modelling approach is adopted, incorporating an analytical and semi-empirical friction law inferred from global dynamometer data to predict local torque contributions based on contact pressure, temperature, and sliding speed. The proposed thermoelastic 3D FVM model was validated against full-scale experimental dynamometer data for standard and wedge-amplified Formula One callipers up to 40° wedge angles. Across diverse operating conditions, including mapping, heavy duty, high torque, and telemetry cycles, the model demonstrated high predictive accuracy. Braking torque predictions yielded normalised peak torque and normalised RMSE errors below 5% for most of the configurations and cycles analysed, while disc surface temperature estimations maintained a RMSE below 50 °C in all conditions. The validated model is applied to predict the performance of a new and untested prototype, designed to deliver the same theoretical braking torque at the same hydraulic pressure as an existing reference braking system. The agreement with both the simulated and experimental results of the reference calliper further suggests the capability of the model as a flexible tool to evaluate the braking performance of early-stage designs, based on their geometry, hydraulic layout, and wedge angle. To assess component impact on overall track performance, the research with Brembo Racing was extended to a vehicle-level co-simulation framework coupling VI-CarRealTime with MATLAB/Simulink. This integration overcame native software limitations by enabling dynamic brake balance management and accurate disc temperature tracking over complete race stints, with the majority of temperature percentage errors between 3% and 8%. Recognising the absence of local friction data for racing-specific C/C composites, a custom pin-on-disc tribometer capable of handling contact pressures up to 5 MPa was mechanically designed and assembled during a joint collaboration with Cranfield University. This hardware lays the essential groundwork for future experimental characterisation of C/C tribology, which will ultimately replace the inferred analytical local friction law to possibly further refine the Dymola model’s predictive fidelity.| File | Dimensione | Formato | |
|---|---|---|---|
|
PhD_Thesis_Draft_Revised_Galvanini.pdf
non disponibili
Licenza:
Tutti i diritti riservati
Dimensione
102 MB
Formato
Adobe PDF
|
102 MB | Adobe PDF |
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.14242/384801
URN:NBN:IT:POLIMI-384801