MulT_predict - A multicomponent geothermometer with integrated sensitivity analyses

For a successful geothermal reservoir exploration, an in-situ temperature estimation is essential. Since geothermometric reservoir temperature estimations using conventional solute geothermometers often entail high uncertainties, a new computational approach is proposed. The goal was to obtain high-accuracy multicomponent reservoir temperature estimations by only using standard geochemical data without the need of sophisticated gas analysis. Therefore, the new numerical tool MulT_predict is introduced. MulT_predict is a multicomponent geothermometer code with integrated sensitivity analyses to back calculate on in-situ conditions. The script is based on MATLAB, which interacts with IPhreeqc. The tool was calibrated and validated against in-situ reservoir temperature measurements. Hence, reservoir conditions are numerically reconstructed by varying various sensitive parameters (e.g. pH value, steam loss, aluminum concentration etc.) to reduce the uncertainties of the reservoir temperature estimation. The new method led to statistically robust and precise reservoir temperature estimations.
At first, a set of reservoir specific minerals is selected as the base of the multicomponent geothermometry. ... mehrWhile calculating the saturations indices of the mineral phases over a defined temperature range, sensitive parameters are subsequently varied. As pH, aluminum concentration and redox potential are prone to interferences (e.g. measurement errors, secondary processes, etc.) as well as possible phase segregation due to boiling during the fluid ascent, reservoir conditions are numerically reconstructed to reduce the temperature estimation uncertainties. The variation of sensitive parameters minimizes the spread between the calculated temperature estimations of each selected mineral phase. The minimal range within the temperature estimations reflects the most plausible reservoir conditions. In this case, the geochemical equilibrium between mineral phases and the reservoir rock is reconstructed. The reservoir temperature estimations fitting the in-situ temperature measurements with a maximal uncertainty of 2.6% and an overall temperature accuracy of 0.5% while the average temperature spread is about 4.2% of the measured absolute reservoir temperature. Furthermore, the back calculated sensitive parameters match the results corrected via WATCH 2.4 (Bjarnason 2010). Especially steam loss and pH value now can be reconstructed with just a standard water analysis without the requirement for an additional gas analysis other approaches typically need. In addition, no supplementary software is needed to back calculate nor pH value nor steam loss. The outcome of the statistical evaluation is given as a box plot combining the temperature estimations of each mineral phase used in the multicomponent geothermometer. In conclusion, the developed method is a promising tool for the high precision estimation of reservoir temperatures. Since MulT_predict does not rely on a sophisticated gas analysis and geochemical data, which is often not available, the tool facilitates the usability yet calculating precise reservoir temperature estimations.