Συντάχθηκε 08-10-2015 13:26
από Styliani Raka
Email συντάκτη: sraka<στο>tuc.gr
Ενημερώθηκε:
-
Κύρια: ΕΤΕΠ ΜΗΧΟΠ.
Άλλες ιδιότητες: απόφοιτος προπτυχιακός ΜΗΧΟΠ, απόφοιτος ΜΔΕ/Διδ. ΜΗΧΟΠ
ΠΟΛΥΤΕΧΝΕΙΟ ΚΡΗΤΗΣ
Σχολή Μηχανικών Ορυκτών Πόρων
Παρουσίαση Μεταπτυχιακής Εργασίας
Εξεταζόμενος φοιτητής: Παπαδάκης Δημήτριος
Θέμα Μεταπτυχιακής Εργασίας: «Family affiliation of Devonian Western Canada oils, using chemometric methods»
Ημέρα παρουσίασης: Τετάρτη, 14 Οκτωβρίου 2015
Ώρα παρουσίασης: 13:30
Τόπος παρουσίασης: Αίθουσα Μεταπτυχιακού Προγράμματος «Μηχανική Πετρελαίου» Μ3.003
Τριμελής Εξεταστική Επιτροπή
Καθ. Ν. Πασαδάκης (επιβλέπων)
Καθ. Δ. Χριστόπουλος
Καθ. Ν. Βαρότσης
Περίληψη
The gasoline range and saturate fractions compositional data of oils carry significant geochemical information. In this master thesis we investigate the ability of chemometric methods in revealing of the oil families for the Devonian petroleum systems. The multivariate statistical methods were applied to the gasoline range and saturate fraction data for 146 oil samples from the Western Canada Sedimentary Basin.
Master thesis is organized in four chapters. Chapter 1 focuses on the introduction of the geological setting of the sampling area. Chapter 2 covers the principles for the methods of dimensionality reduction. The methods employed here are: hierarchical clustering, principal component analysis (PCA), kernel principal component analysis (KPCA), k-means clustering and hierarchical clustering. In chapter 3 we focus in a more detailed examination of geochemical information contained in gasoline and saturated fractions and we introduce the characteristics indices for these fractions. Finally in chapter 4 we examine four models aiming to separate the oils into families and we investigate the performance of each model in family affiliation. For the first two models normalized values of original component data for the gasoline and saturated fractions were used, while in models 3 and 4 calculated characteristic geochemical indices were used. Due to geochemical similarity among oil samples, the first two models did not give satisfactory results for oil affiliation. Model 3 gives the best results and reveals two distinct families in oil samples and finally model 4 give results for three different oil families.
Από τη Γραμματεία της Σχολής ΜΗΧΟΠ
