Modelling of tracer tests in a geothermal reservoir in Tianjin, China

Abstract

The Wumishan (Jxw) geothermal reservoir in the Dongli Lake area is an extensive, low-temperature geothermal system hosted mainly by Mesoproterozoic dolomitic limestone. In order to study the flow paths and predict the cooling due to long term injection, tracer tests were performed on the 17th of December, 2015. 700 kg of Ammonium Molybdate (Mo) were injected into well DL-48B. No obvious tracer recovery was detected in the water samples collected during the 90 days of tracer test. In order to interpret the tracer testing quantitatively, both an analytical method developed by ISOR (Iceland GeoSurvey) and a numerical method using Visual MODFLOW flex software were applied. For the analytical model, it was assumed that the recovery is very slow and would appear after 90 days. In the three cases of longitudinal dispersivity, 77, 230 and 384 m, the tracer concentration began to increase after 150 days. When the dispersivity was 77 m, thermal breakthrough would occur in 80 years for a narrow flow channel scenario and 27 years for a wide flow channel scenario with an average annual production and injection rate of 10 kg/s. Using an automatic parameter estimation tool along with the numerical model, errors were minimized between the observed and the simulated data to estimate the distribution of the reservoir parameters. There is reasonable agreement between the simulated and observed water levels and therefore the numerical model was used to predict the tracer concentration in the production well. For the most pessimistic case (longitudinal dispersivity of 384 m), the tracer will take more than a year to arrive at the production well with concentration values outside of the detection limit, according to the model. Results show that there is no direct connection between production and injection wells and neither of the models predicts cooling in the next 24 years.