Knock is one of the main factors limiting the performance of high-performance spark-ignition engines. Reliable knock prediction during the development phase requires an accurate three-dimensional representation of both combustion and charge autoignition processes. This thesis presents the development and validation of a 3D CFD model for knock simulation in a high-performance racing engine, from the definition of the combustion chemistry to its implementation and calibration within the engine model. Combustion is described through a hybrid approach aimed at limiting the computational cost: flame-front propagation is modelled using the Turbulent Flame Speed Closure (TFC) model, while the chemistry of the burned gases and end-gas is solved explicitly through the Complex Chemistry model implemented in Simcenter STAR-CCM+. This approach aims to describe knock onset directly through the chemical kinetics of the end-gas, without relying on calibrated empirical correlations. The main contribution of this work is the development and validation of the chemical-kinetic modelling framework required to make this approach applicable. A surrogate fuel representative of the real racing fuel was formulated through a multi-objective optimization procedure, and a detailed kinetic mechanism was validated against the main combustion properties, namely ignition delay times and laminar flame speeds. Starting from this mechanism, two complementary reduced mechanisms were derived through automated reduction procedures: one dedicated to chemistry resolution within the CFD domain and the other to the generation of the laminar flame speed tables employed by the TFC model. The mechanism intended for resolved chemistry was further refined through calibration of the Arrhenius parameters of selected elementary reactions in order to improve heat-release-rate prediction while preserving accuracy in the description of autoignition. The reduced mechanism and the laminar flame speed tables were finally integrated into the in-cylinder 3D CFD engine model, where the TFC model was calibrated against the available experimental data until the combustion phasing was reproduced with heat-release-rate amplitude errors on the order of one percent. The resulting model correctly predicts the increase in knock propensity produced by advanced spark timing but does not reproduce end-gas autoignition. The comparison between the Livengood–Wu integral criterion and the resolved chemistry, together with the analysis of intermediate autoignition species, narrows down the possible causes of this discrepancy and identifies the temporal resolution adopted for kinetic integration as the most likely explanation, although the available evidence does not allow alternative hypotheses to be conclusively ruled out. The work concludes by outlining the investigations required to discriminate among these possible causes.
Il knock rappresenta uno dei principali fattori che limitano le prestazioni dei motori ad accensione comandata ad alte prestazioni. La sua previsione affidabile in fase di sviluppo richiede una rappresentazione tridimensionale accurata sia della combustione sia dei processi di autoaccensione della carica. Questa tesi presenta lo sviluppo e la validazione di un modello CFD-3D per la simulazione del knock in un motore da competizione, dalla definizione della chimica di combustione fino alla sua implementazione e calibrazione nel modello motore. La combustione è descritta mediante un approccio ibrido volto a contenere il costo computazionale: la propagazione del fronte di fiamma è modellata tramite il modello Turbulent Flame Speed Closure (TFC), mentre la chimica dei gas combusti e dell'end-gas è risolta esplicitamente attraverso il modello Complex Chemistry implementato in Simcenter STAR-CCM+. Tale approccio ha l'obiettivo di descrivere l'insorgenza del knock direttamente attraverso la cinetica chimica dell'end-gas, senza ricorrere a correlazioni empiriche calibrate. Il contributo principale del lavoro riguarda lo sviluppo e la validazione della catena di modellazione cinetico-chimica necessaria a rendere applicabile questo approccio. È stato formulato un combustibile surrogato rappresentativo del carburante reale mediante ottimizzazione multi-obiettivo ed è stato validato un meccanismo cinetico dettagliato rispetto alle principali proprietà di combustione, quali tempi di ritardo all'autoaccensione e velocità di fiamma laminare. A partire da esso sono stati derivati, tramite procedure di riduzione automatizzata, due meccanismi compatti complementari: uno dedicato alla risoluzione della chimica nel dominio CFD e uno alla generazione delle tabelle di velocità di fiamma laminare impiegate dal modello TFC. Il meccanismo destinato alla chimica risolta è stato inoltre affinato mediante la calibrazione dei parametri di Arrhenius delle reazioni elementari, al fine di migliorare la riproduzione del rilascio di calore mantenendo l'accuratezza nella descrizione dell'autoaccensione. Il meccanismo ridotto e le tabelle di velocità di fiamma laminare sono stati infine integrati nel modello CFD-3D in-cylinder del motore, dove il TFC è stato calibrato sui dati sperimentali disponibili fino a riprodurre la fasatura della combustione con scarti dell'ordine dell'uno per cento sulle ampiezze di rilascio del calore. Il modello così ottenuto riconosce correttamente l'aumento di propensione al knock prodotto da un anticipo di accensione maggiorato, ma non risolve l'evento di autoaccensione dell'end-gas. Il confronto tra il criterio integrale e la chimica risolta, e l'esame delle specie intermedie dell'autoaccensione, circoscrivono le cause possibili di questa mancanza e indicano nella risoluzione temporale con cui la cinetica viene integrata l'ipotesi più probabile, senza che i dati disponibili consentano di escluderne altre. Il lavoro si chiude individuando le verifiche necessarie a discriminarle.
Calibrazione e Validazione di un Modello di Knock CFD-3D In-Cylinder per Motori ad Accensione Comandata
LORENZI, NICOLO'
2025/2026
Abstract
Knock is one of the main factors limiting the performance of high-performance spark-ignition engines. Reliable knock prediction during the development phase requires an accurate three-dimensional representation of both combustion and charge autoignition processes. This thesis presents the development and validation of a 3D CFD model for knock simulation in a high-performance racing engine, from the definition of the combustion chemistry to its implementation and calibration within the engine model. Combustion is described through a hybrid approach aimed at limiting the computational cost: flame-front propagation is modelled using the Turbulent Flame Speed Closure (TFC) model, while the chemistry of the burned gases and end-gas is solved explicitly through the Complex Chemistry model implemented in Simcenter STAR-CCM+. This approach aims to describe knock onset directly through the chemical kinetics of the end-gas, without relying on calibrated empirical correlations. The main contribution of this work is the development and validation of the chemical-kinetic modelling framework required to make this approach applicable. A surrogate fuel representative of the real racing fuel was formulated through a multi-objective optimization procedure, and a detailed kinetic mechanism was validated against the main combustion properties, namely ignition delay times and laminar flame speeds. Starting from this mechanism, two complementary reduced mechanisms were derived through automated reduction procedures: one dedicated to chemistry resolution within the CFD domain and the other to the generation of the laminar flame speed tables employed by the TFC model. The mechanism intended for resolved chemistry was further refined through calibration of the Arrhenius parameters of selected elementary reactions in order to improve heat-release-rate prediction while preserving accuracy in the description of autoignition. The reduced mechanism and the laminar flame speed tables were finally integrated into the in-cylinder 3D CFD engine model, where the TFC model was calibrated against the available experimental data until the combustion phasing was reproduced with heat-release-rate amplitude errors on the order of one percent. The resulting model correctly predicts the increase in knock propensity produced by advanced spark timing but does not reproduce end-gas autoignition. The comparison between the Livengood–Wu integral criterion and the resolved chemistry, together with the analysis of intermediate autoignition species, narrows down the possible causes of this discrepancy and identifies the temporal resolution adopted for kinetic integration as the most likely explanation, although the available evidence does not allow alternative hypotheses to be conclusively ruled out. The work concludes by outlining the investigations required to discriminate among these possible causes.| File | Dimensione | Formato | |
|---|---|---|---|
|
Lorenzi.Nicolò.pdf
embargo fino al 21/07/2029
Dimensione
10 MB
Formato
Adobe PDF
|
10 MB | Adobe PDF |
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.14251/7577