Forecasting high resolution variations in deep geothermal wells based on low resolution measurements utilizing virtual sensors

Abstract

<p>Multiple operators use geothermal wells in Lower Bavaria and Upper Austria for balneological (medical and wellness) applications and for energy and/or heat mining purposes. These applications depend heavily on the well‘s hydrochemical and geophysical stability (mineralization, pressure, temperature).</p><p>At the moment, wells are submitted to inspection once a year, which includes the analysis of ions, pH, pressure, temperature etc. These „offline“ analyses, while covering a large set of parameters, obviously fail to show intra-annual variability within the measured parameters. On the other hand, some geothermal wells are being monitored quasi-continually. These „online“ sensors, however, only cover a small set of selected parameters, such as electric conductivity, temperature and pressure.</p><p>This study aims at forecasting hydrochemical and physical stability based on annual measurements by assessing the degree of intra-annual variability covered or neglected by the yearly measurements. The results are required for a sound assessment of possible adverse effects of other exploration activities and short term variations of the withdrawal rates reflecting the demand for heat, energy and/or spa water. To do this, we followed the concept of virtual sensors and their correlation to detailed yearly measurements.</p><p>We found that, while annual measurements, when taken approximately in the same season of the year, do match the data sampled online quite well, intra-annual variability at the examined wells was quite strong for some parameters and not represented by the offline data. Thus, annual data can be used to make predictions regarding long-term variability. In order to forecast intra-annual variability, higher temporal resolution is necessary. While not a replacement for the detailed analyses, the virtual sensors presented here provide a robust method to trigger further actions.</p>