The overall aim of this thesis was the identification and quantification of the hydraulic and hydrogeochemical key processes controlling the water quality changes during a reclaimed water ASR experiment in Bolivar and during artificial recharge at an infiltration pond in Berlin. This aim was successfully met with a variety of field and modelling methods. Based on the results of Chapters 2, 3 and 4, this chapter summarises the major conclusions of this research and gives prospects for future research on the investigated artificial recharge systems and reactive transport modelling of such systems in general.
Since in this thesis two different artificial recharge schemes in completely different field situations were investigated, the conclusions are predominantly site-specific, and thus are presented here separately for each field site. Conclusions relating to the reactive transport modelling issues are given thereafter.
A quantitative process-based modelling framework was developed and applied to an ASR experiment at Bolivar, Australia, in order to simulate the geochemical response during injection, storage and recovery. The developed modelling framework is capable of providing a very good quantitative description of the physical and biogeochemical processes that occurred during the injection, storage and recovery periods. The modelling study in particular demonstrates that the explicit incorporation of bacterial mass is essential to describe the local hydrogeochemical effects occurring in the vicinity of the injection/extraction well during the storage phase. Without a detailed microbial growth and decay model, these hydrogeochemical observations could not be reproduced at all (see Appendix A). The model shows that hydrogeochemical changes further away from the injection/extraction well were mainly driven by ion exchange and calcite dissolution. Microbial mass dynamics have no relevance for the hydrochemical changes further away from the injection/extraction well. In summary, the model provides a consistent interpretation of how different biogeochemical processes affect the injected plume on its full scale. Besides the advanced process understanding, a practical aspect of this study lies in the preliminary evaluation of how much the aquifer properties might by altered by physical, chemical and biological clogging processes. The model simulations affirm that significant local conductivity reduction did not occur during the field study.
The artificial recharge pond in Berlin was investigated in Chapter 3 and 4 of this thesis with the aims of (i) identifying the hydraulic regime and its impact on the dynamics of the hydrogeochemical environment immediately below the pond, and (ii) providing a quantitative process-based modelling framework for simulating the fate of the redox-sensitive PhAC phenazone within the surrounding aquifer of the pond.
Directly below the recharge pond, the hydrogeochemical environment is predominantly impacted by transient hydraulic conditions and seasonal temperature variations. Thereby, the entire hydraulic system is controlled by the formation of a clogging layer at the bottom of the pond resulting in alternating saturated/unsaturated conditions below the pond. During summer, when the infiltrated water is relatively warm (about 20-25°C), the spatial and temporal distribution of different redox conditions is strongly determined by the prevailing hydraulic conditions and their dynamics below the pond. When the sediment below the pond is fully water saturated in the earlier part of the operational recharge cycle, nitrate reducing and manganese reducing conditions are dominant. Iron and sulphate reducing conditions develop only in anaerobic micro-sites due to chemical and/or physical heterogeneity. When the hydraulic conditions change from water saturated to water unsaturated conditions, which is approximately in the middle of the recharge period, remarkable changes of the local hydrogeochemistry below the pond can be observed. They result from the temporal re-oxidation of previously formed iron-sulphides due to the lateral intrusion of atmospheric oxygen penetrating from the pond margins into the centre below the pond. During the following unsaturated period, the availability of atmospheric oxygen enhances the mineralisation of particulate organic carbon, which in turn promotes the dissolution of calcite. Whereas in summer the dynamics and the character of the hydrochemical system are strongly affected by the hydraulic situation below the pond, hydrogeochemical changes are relatively independent from the hydraulic conditions in winter. During the entire winter period, the redox status below the pond remains aerobic due to decreased biodegradation rates at lower temperatures (see Appendix B). Since the redox system does not shift to nitrate reducing conditions during winter, nitrification of sedimentary bound nitrogen can be observed as elevated nitrate concentrations in the groundwater. This could indicate that nitrification below the pond is less affected by temperature changes than nitrate reduction.
The modelling study for this site showed that at larger scale, the hydrogeochemical processes, in particular the redox processes, are predominantly controlled by seasonal temperature changes rather than transient groundwater flow due to varying recharge rates. The influence of transient saturated/unsaturated conditions appeared not to be important for the behaviour of the hydrogeochemical system on that scale. It could be clearly demonstrated that solely the redox dependency of the degradation rate of phenazone controls the fate of phenazone in the subsurface of this artificial recharge site.
Throughout the modelling procedures in Chapter 2 and 4, the capabilities of mechanistic multi-component reactive transport modelling for interpreting field observations and understanding and quantifying the non-intuitive interactions of processes in artificial recharge systems were intensively tested. Both modelling studies clearly illustrate the suitability of multi-component reactive transport modelling as a fundamental tool for analysing the complex hydrogeochemical processes occurring in artificial recharge systems. It was demonstrated that such models could be used to successfully evaluate the underlying physical, chemical and biological processes separately, which in reality are highly co-dependent. This allows a systematic development and verification of new hypotheses of how these processes interact. Furthermore, the application of the model independent parameter estimation program PEST to the Bolivar reactive transport model was found to be particular useful, since the automatically generated calibration statistics helped to evaluate the selected conceptual model and its parameterisation. However, the processes and the respective parameters that were identified for both artificial recharge systems are, of course, subject to the ever-present uncertainty problem. Thus, it’s clear that there might be other conceptual models, which describe the field observations equally well. An extensive predictive capability of the developed models at this stage is therefore unrealistic, especially for other field sites with different recharge water and site characteristics. However, future studies for other sites or long-term data of the Bolivar and Berlin site will provide additional constrains to test the present and potential alternative conceptual models. Fortunately, the flexible nature of sophisticated and user-friendly modelling tools such as PHT3D allow a quick and steady improvement and adoption of conceptual models rather than to focus on a single conceptual model where pre-defined parameters are fitted.
As outlined in the previous section, for a further verification of the present conceptual models developed in Chapter 2 and 4, it would be necessary to extend the simulation periods where data is available. This is especially needed for the Bolivar site, as it has not been tested yet if the model is able to adequately reproduce the geochemical evolution of the injected plume, the recovered water and the groundwater for more than the one ASR-cycle. After an extension of the simulation period and a possible re-calibration of the reactive transport model for the Bolivar site, the incorporation of disinfection-by-products as components into the modelling framework might be of interest, since it has been found that the fate of some of these compounds are highly dependent on the local geochemical environment (Pavelic et al., 2005). Analogue to the investigation of the fate of phenazone at the recharge site in Berlin, such an extended model can help to find and quantify the key processes of the attenuation behaviour of these compounds during ASR.
Since the monitoring of hydrogeochemical changes within the unsaturated zone below the recharge pond at the Berlin site was restricted to the fringe area (Chapter 3), the magnitude of hydrochemical changes of the seepage water in the centre area of the pond is still unclear. In order to fill this gap, future investigations on that site are recommended including (i) an extended field campaign monitoring the relevant hydraulic and hydrochemical parameters in the entire cross-sectional area below the pond, and (ii) the development of a two-dimensional multi-component reactive transport model for this cross section, considering advective-dispersive transport of O2 and CO2 in the gas phase and major ions in the water phase during the unsaturated stage.
With the successful simulation of the redox dynamics and the fate of phenazone at the Berlin site, one can expect that it is also possible to model the fate of other phenazone-type pharmaceuticals such as 1-acetyl-1-methyl-2-dimethyloxamoyl-2-phenylhydrazide (AMDOPH), 1-acetyl-1-methyl-2-phenylhydrazid (AMPH), Acetoaminoantipyrin (AAA), formylaminoantipyrin (FAA), 1,5-dimethyl-1,2-dehydro-3-pyrazolone (DP) and 4-(2-methylethyl)-1,5-dimethyl-1,2-dehydro-3-pyrazole (PDP) that were detected at the artificial recharge site in Berlin (Massmann et al., 2005b).
An appealing option for the site operator (BWB) might be the ability to simulate the quality of the water extracted by the surrounding production well field, because the extraction water quality is likely to be altered by operational changes of the pumping regime or recharge rates. The development of a three-dimensional reactive transport model for a larger model domain capturing the production well field is undoubtedly possible, since a calibrated site-specific reaction module for the aquifer now exists (Chapter 4).
More general, for future investigations on artificial recharge systems I would strongly recommend the integration of mechanistic reactive transport modelling into the working program. This would be especially beneficial when looking at potential harmful compounds or pathogens, as multi-component reactive transport models intrinsically determine the local geochemical environment such as major ion concentration, ionic strength, pH and redox potential, which the fate of these contaminants often depends on. Moreover, as exemplarily demonstrated in Chapter 2, using automatic calibration procedures in conjunction with the development of conceptual reaction models can give valuable insights into the suitability of the chosen parameterisation. This means, for a systematic evaluation of the developed conceptual models, there is an urgent need for integrating automatic parameter estimation into the field of reactive transport modelling.
|© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.|
|DiML DTD Version 4.0||Zertifizierter Dokumentenserver|
der Humboldt-Universität zu Berlin
|HTML-Version erstellt am: