Abstract
Forecasting of downhole temperatures is of great interest for the development of deep geothermal energy, since the maximum production temperatures are important for the efficiency of the system. The production temperatures are mainly determined by the prevailing reservoir temperature, which is insufficiently known even in well-developed reservoirs. Most available temperature data from hydrothermal reservoirs are “bottom hole temperatures” (BHTs) that are usually measured during geophysical measurement programs, after each section of a deep geothermal well has been drilled. These measured temperatures are thermally disturbed by the preceding drilling fluid circulation and therefore show a high deviation from the undisturbed formation temperature, requiring correction of the BHT measurements. This is made possible by a variety of analytical and numerical BHT correction methods, all of which require different input parameters for each method. Those parameters are often documented with poor quality, incompletely, or not at all. It can be assumed that the inaccuracy of a corrected BHT value depends to a high degree on the errors of the input parameters and the choice of the correction method is secondary in this respect.In order to perform a complete evaluation of corrected BHT values and to determine the range of errors, we corrected BHT values from over 300 current geothermal and old hydrocarbon wells in the Bavarian Molasse Basin in Southern Germany using a Monte Carlo approach. Thus, the corrected temperature was given as a density distribution rather than a discrete value after individual estimation of the error in the input parameters. This allows a prediction of the formation temperature based on risk scenarios, for example specifying a p10 or p90 case.From the corrected temperatures and taking into account the individual variances studied (p10 and p90 values); we created a set of temperature gradients that take into consideration, if known, the discovered inflow zones and the slope changes in the stratigraphic layers. This approach provides a spatial representation of temperatures while accounting for error propagation by estimates in the correction process, as well as the extrapolation of point temperature data to gradients and the application of geo-statistical methods.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have