BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//TUC//Events//EN
CALSCALE:GREGORIAN
BEGIN:VTIMEZONE
TZID:Europe/Athens
TZNAME:EEST
DTSTART:19700329T030000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0300
TZNAME:EET
DTSTART:19701025T040000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CREATED:20260706T124216Z
LAST-MODIFIED:20260706T124216Z
DTSTAMP:20260717T014951Z
UID:1784242191@tuc.gr
SUMMARY:Παρουσίαση ΜΔΕ- ΣΠΗΛΙΟΠΟΥΛΟΣ ΓΡΗΓΟΡΙ
 ΟΣ -Σχολή ΧΗΜΗΠΕΡ
LOCATION:Κ2 - Κτίριο ΧΗΜΗΠΕΡ
DESCRIPTION:https://www.tuc.gr/el/to-polytechnei
 o/ilektronikes-ypiresies/imerologio/
 imerologio-ekdiloseon-1?tx_tucevents
 2_tuceventsdisplay%5Baction%5D=show&
 tx_tucevents2_tuceventsdisplay%5Bcon
 troller%5D=Event&tx_tucevents2_tucev
 entsdisplay%5Bevent%5D=8542&cHash=86
 083f12d4b40f2f23206ba40152f25e\nANNO
 UNCEMENT OF PRESENTATION OF POSTGRAD
 UATE THESIS\n \n First Name/Surname:
  Grigorios Spiliopoulos\n Student Id
 entification Number: 2024057019\n Da
 te: 08/07/2026\n Time: 10:00 (GMT +3
 )\n Room / Zoom Link: https://tucgr.
 zoom.us/j/91477145205?pwd=F3TMap3Sob
 Wa25XZHko4P8O8kla93b.1  \n Meeting I
 D: 914 7714 5205\n Password: 878737\
 n \n Title: “Algorithmic Synthesis a
 nd Extrapolation of Wind Tunnel Aero
 dynamic Coefficients for Single-Axis
  Tracker Systems”\n \n Supervisor: P
 rof. Theocharis Tsoutsos\n Three-mem
 ber committee:\n 1. Prof. Theocharis
  Tsoutsos\n 2. Prof. Apostolos Voulg
 arakis\n 3. Assoc. Prof. Alexandros 
 Stefanakis \n \n Abstract:\n The rel
 iability of the structural integrity
  of single-axis photovoltaic (PV) tr
 acking systems at a utilityscale, ar
 e highly dependent upon accurate aer
 odynamic pressure coefficients deter
 mined through \n wind tunnel testing
 . The physical scaling limitations p
 resented by the prototype and the ec
 onomy of \n testing, restrict the ab
 ility to conduct wind tunnel tests f
 or each row pitch, ground clearance 
 and tilt \n angle. This, results in 
 critical gaps in the existing aerody
 namic pressure coefficient data. Thi
 s thesis\n presents a computational 
 framework, referred to as WindFit, t
 hat provides the means to synthesize
  \n the transient aerodynamic loads 
 for the geometries of PV trackers th
 at have not been tested.\n The metho
 dology employs localized polynomial 
 regression to reconstruct the incomp
 lete azimuth and \n tilt data associ
 ated with tracker operation. Geometr
 ic synthesis will be achieved by pro
 jecting \n structural parametric dat
 a onto a Cartesian coordinate system
 . Geometries that fall into an empir
 ical \n convex hull will be resolved
  using Barycentric Coordinate Mathem
 atics, while those that fall outside
  \n the convex hull, will utilize th
 e Virtual Anchor Vertex protocol wit
 hin the framework, to constrain \n t
 he extrapolated data. This topology-
 constrained boundary gradient approa
 ch prevents the numerical \n diverge
 nce and reversed aerodynamic weighti
 ng that are associated with traditio
 nal unconstrained \n extrapolation m
 ethods.\n The computational effectiv
 eness of the framework is verified b
 y two benchmark approaches, which \n
  use deterministic Inverse Distance 
 Weighting (IDW) and stochastic Gauss
 ian Process Regression \n (GPR) mode
 ls. The comparative spatial validati
 on of the models shows that IDW oper
 ates as a low \n pass filter, creati
 ng spatial blurring that greatly und
 er-predicts the high-end boundary lo
 ads, as well \n as the Venturi ampli
 fications. The GPR model produces th
 e same high-fidelity precision as Wi
 ndFit, \n but due to it's reliance o
 n infinite mathematical smoothness, 
 GPR removes sharp aerodynamic \n dis
 continuities and dramatically under-
 predicts detached tip vortexes durin
 g extrapolation. \n Finally, WindFit
  integrates local geometric interpol
 ation with boundary conditions based
  upon \n physics to accurately model
  complex flow separation phenomena. 
 In addition, it uses Barycentric \n 
 algebraic weights to eliminate the e
 xtensive latency involved with perfo
 rming machine learning \n matrix inv
 ersions, thus making this approach h
 ighly suited for real-time parametri
 c structural design\n
STATUS:CONFIRMED
ORGANIZER;RSVP=FALSE;CN=TUC;CUTYPE=TUC:mailto:webmaster@tuc.gr
DTSTART:20260708T100000
DTEND:20260708T110000
TRANSP:OPAQUE
CLASS:DEFAULT
END:VEVENT
END:VCALENDAR