This thesis has been developed thanks to the collaboration between the Politecnico of Milano and SDF spa. The cooling system of a tractor is the object of the study. In particular, its main variables are statistically analyzed, meaning the research of their tolerances and the statistical distributions of the heat exchanger performances, heat rejection, geometrical characteristics and fan performance. These collection of data is then used in a design of experiment (DOE). This instrument, by means of CFD simulations and the Monte Carlo analysis, is able to generate the statistical behavior of the overall cooling system performances, i.e. the coolant air to boil (ATB), the fan power and the temperatures at the charge air and oil cooler. The project is able to generate a classification of the main influent cooling system variables, showing the actual percentage of tractor that statistically are not respecting SDF spa targets and possibly answering to past CFD and experimental result difference. After a screening phase, the heat exchanger performances are not included in the DOE, as well the fan deformation and slip factor variation, which are treated as future developments. Considering the geometrical cooling system characteristics and the heat rejection as inputs of the DOE, it turns out that the most relevant variables are mainly four. The axial and vertical position of the fan, with respect to the shroud, together with the diameter of the fan, because of their strong impact on the cooling air fluxes, and the thermal power exchanged at the water cooler. The Monte Carlo analysis generates 99% confidence intervals of the above mentioned monitored cooling system performances. In particular, the ATB is considered the ID of the cooling system and it shows a tolerance of 2.76°C with a standard deviation equal to 0.46. Due to confidential reasons, it is not possible to report the percentage of the SDF spa targets outside these intervals. The possible mismatching between CFD simulations and experimental tests is proved as well.
Questa tesi è stata sviluppata grazie alla collaborazione tra il Politecnico di Milano e SDF spa. Il sistema di raffreddamento di un trattore è l’oggetto dello studio. In particolare, le principali variabili sono analizzate statisticamente, effettuando una ricerca delle tolleranze e distribuzioni aleatorie delle presentazioni degli scambiatori di calore, delle potenze termiche scambiate, delle caratteristiche geometriche e delle prestazioni del ventilatore. Questa raccolta di dati è poi utilizzata in un design of experiment (DOE). Questo strumento, attraverso simulazioni CFD e un’analisi di Monte Carlo, ricava il comportamento statistico delle prestazioni finali del sistema di raffreddamento, quali l’air-to-boil (ATB), la potenza assorbita dal ventilatore e le temperature dello scambiatore dell’olio e dell’aria compressa. Il progetto è in grado di generare una classificazione delle variabili più influenti, mostrando l’attuale percentuale di trattori che non rispetta i target aziendali, e spiegare la possibile differenza tra i test CFD e sperimentali riscontrata in passato. Dopo una preselezione, le prestazioni degli scambiatori di calore non sono state incluse nel DOE, così come la deformazione del ventilatore e il suo coefficiente di attrito, i quali sono trattati come sviluppi futuri del progetto. Considerando le caratteristiche geometriche e le potenze scambiate come input del DOE, quattro variabili risultano essere particolarmente influenti: la posizione assiale e verticale del ventilatore, rispetto al convogliatore, e il diametro delle pale, a causa della forte azione sui flussi dell’aria, così come la potenza scambiata al radiatore dell’acqua. L’analisi di Monte Carlo genera intervalli di confidenza al 99% per ciascun target menzionato in precedenza. In particolare, l’ATB è considerato l’indice principale e mostra una tolleranza di 2.76°C, con una deviazione standard di 0.46. Per ragioni confidenziali, non è possibile riportare la percentuale attuale dei target SDF spa al di fuori di questi intervalli. Infine, il progetto giustifica la possibile differenza tra simulazioni CFD e test sperimentali.
Advanced statistical analysis of a tractor cooling system
SILVA, RICCARDO
2017/2018
Abstract
This thesis has been developed thanks to the collaboration between the Politecnico of Milano and SDF spa. The cooling system of a tractor is the object of the study. In particular, its main variables are statistically analyzed, meaning the research of their tolerances and the statistical distributions of the heat exchanger performances, heat rejection, geometrical characteristics and fan performance. These collection of data is then used in a design of experiment (DOE). This instrument, by means of CFD simulations and the Monte Carlo analysis, is able to generate the statistical behavior of the overall cooling system performances, i.e. the coolant air to boil (ATB), the fan power and the temperatures at the charge air and oil cooler. The project is able to generate a classification of the main influent cooling system variables, showing the actual percentage of tractor that statistically are not respecting SDF spa targets and possibly answering to past CFD and experimental result difference. After a screening phase, the heat exchanger performances are not included in the DOE, as well the fan deformation and slip factor variation, which are treated as future developments. Considering the geometrical cooling system characteristics and the heat rejection as inputs of the DOE, it turns out that the most relevant variables are mainly four. The axial and vertical position of the fan, with respect to the shroud, together with the diameter of the fan, because of their strong impact on the cooling air fluxes, and the thermal power exchanged at the water cooler. The Monte Carlo analysis generates 99% confidence intervals of the above mentioned monitored cooling system performances. In particular, the ATB is considered the ID of the cooling system and it shows a tolerance of 2.76°C with a standard deviation equal to 0.46. Due to confidential reasons, it is not possible to report the percentage of the SDF spa targets outside these intervals. The possible mismatching between CFD simulations and experimental tests is proved as well.File | Dimensione | Formato | |
---|---|---|---|
TESI.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: Testo della tesi
Dimensione
3.52 MB
Formato
Adobe PDF
|
3.52 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/146345