![Institutional Repository [SANDBOX]](images/dias.logo.png "Institutional Repository [SANDBOX]"){kind=logo}
Institutional Repository [SANDBOX]
Technical University of Crete
Technical University of Crete
EN
|
EL
| URI: | http://purl.tuc.gr/dl/dias/1C0A3A36-016C-4DB0-BF0B-483936548635 | ||
| Year | 2021 | ||
| Type of Item | Diploma Work | ||
| License |
|
||
| Bibliographic Citation | Fotios Spyropoulos, "Modular calibration improving the boundary conditions of a 3D groundwater model (FEFLOW) for the aquifer of Tympaki, Crete, Greece", Diploma Work, School of Chemical and Environmental Engineering, Technical University of Crete, Chania, Greece, 2021 https://doi.org/10.26233/heallink.tuc.90393 | ||
| Appears in Collections | |||
| Relations with other Items |
The following thesis deals with the improvement of an existing modelof groundwater flow in the region of Tympaki, Crete. The main goal is the reduction of the three error parametres: Bias, RMSE and MAE in comparison to the initial results of the model, in order to increase the descriptive capacity of the model. This is achieved by implementing trends in the boundary conditions which are described by harmonic equations, as they were formed in the initial phase of the model, by two different approaches (Mean Average, Nearest Well). Thereafter the harmonic equations were temporally dislocated in regard of the phase at the point of the start of the simulation and the width of one boundary condition was corrected. In the meantime, tests regarding the correlation of precipitation with hydraulic heads were conducted so as to define the reaction time of the system. Evidentially, the most fitting approach was the Nearest Well by improving the Bias by 22,38%, the RMSE 14,88% and MAE 16,08% compared to the initial approach. Whereas by examining the curves of the simulated hydraulic heads before the calibration in comparison with the observed hydraulic head values, a dissonance is apparent. After the presented calibration said dissonance is significantly diminished. A literature review is also presented with the goal of uncovering possible oversights in the initial construction of the model and gaining a general idea for the direction the methodology of the calibration will follow and eventually present proposals for further improvement after the end of the present analysis. The most significant proposals are the use of neural networks in order to produce a metamodel that simulates the boundary conditions and the implementation of complex analytical equations describing boundary conditions.
Copyright © DIAS 2013
