Water level response to harmonic disturbances in confined aquifers: modeling, global sensitivity, and applications

by Yixuan Xing
Doctoral thesis
Date of Examination:2024-06-04
Date of issue:2024-06-13
Advisor:Prof. Dr. Thomas Ptak-Fix
Referee:Prof. Dr. Ekkehard Holzbecher
Referee:Prof. Dr. Rui Hu
Referee:Prof. Dr. Hongbiao Gu
Referee:Dr. Jannes Kordilla
Referee:Dr. Linwei Hu
Persistent Address: http://dx.doi.org/10.53846/goediss-10540

 

 

Files in this item

Name:Xing_theisis_final_without CV.pdf
Size:6.11Mb
Format:PDF

The following license files are associated with this item:

Abstract

English

In well-aquifer system, oscillatory water levels are often observed in monitoring wells when the aquifer is perturbed by harmonic disturbances, such as oscillatory pumping, ocean tide and seismic waves. In general, the water level measured in the wellbore comes from a combination of aquifer disturbance and wellbore effects. Consideration was given to four sources of disturbance, i.e. seismic waves, earth tides, ocean tides and oscillatory pumping. According to the mechanical characterization of the disturbance sources, the sources are classified as periodic pressure diffusion (PD) sources and volumetric strain (VS) sources. Facing different types of disturbance sources, handling the wellbore effect varies. Firstly, a general numerical model integrating different types of harmonic disturbance sources is proposed to investigate the impact of wellbore effects on water level oscillations. Performance of the general numerical model was evaluated by comparing the simulation results with established models for oscillatory pumping tests and seismic waves. The influence of four key parameters was investigated, namely water column height, hydraulic conductivity, disturbance period and distance from the source of disturbance. The essential importance of water column height, disturbance period and distance from the source of disturbance on the wellbore effect is clarified. A global parameter sensitivity of the coupled well-aquifer model is analyzed in the context of the general numerical model. Parameter uncertainties in aquifer properties, source characteristics and wellbore construction were first analyzed for the PD and VS sources, respectively. A Monte Carlo-based surrogate model between the parameters and sensitivity indices was then developed using the Random Forest algorithm. With PD sources, the sensitivity results indicate that hydraulic conductivity is the most sensitive parameter to the yes/no response. When the hydraulic conductivity exceeds 10-4 m/s, the water level observations essentially represent the pore pressure in the aquifer. Noticeable amplitude changes and phase delays occur when the disturbance period is less than 50 seconds, which is mainly due to water column inertia. With VS sources, the water level response generally occurs in aquifers with hydraulic conductivities greater than 10-7 m/s and with a source disturbance with an amplitude greater than 10-3 MPa. Notable amplitude changes and phase shifts, however, occur when the source period varies within 100 seconds. In interpreting the water level response due to all periodic aquifer disturbances, a consideration of the influence of wellbore effects is necessary, especially in low-conductivity aquifers with conductivity less than 10-4 m/s.
Keywords: water level responses; numerical modeling; global sensitivity; random-forest based surrogate model; co-seismic water level interpretation
 
Statistik