In the Onuma geothermal field, brines contain SiO2-concentrations of about 514 mg SiO2 per liter brine. The pH value of the brines is 7.2±0.1 at a temperature of 90°C. The dissolved SiO2 precipitates during flashing of the brines and causes severe problems by forming scales in the reinjection wells. Therefore, an experimental study was designed to explore how SiO2 can be removed from the brines most efficiently and in a controlled way. First, we performed tests with various reagents, i.e., high early-strength portland cement, CaCO3 (calcite), CaO, Ca (OH)2, and CaMg (CO3)2 (dolomite) to explore the efficiency of the reagents in removing silica from the brine. Our test showed that CaO is the most efficient reagent in removing the excess silica from the brine. Therefore, we used CaO and the following method to conduct all subsequent experiments : In a first step, CaO is added continuously to the brine. The CaO-reagent causes the precipitation of amorphous Ca-silicate. Then, in a second step, we reinject the precipitate to fresh brine to cause additional precipitation of Ca-silicate. We conducted all experiments at two different conditions: addition of 0.6g CaO/l and 1.5g CaO/l. The main results are an addition of 0.6g CaO/l reduced the silica concentration to 223 mg/l and increased the pH of the brine to 9.9. An addition of 1.5 g CaO/l caused a significantly stronger reduction in silica concentration of 4 mg/l and raised the pH value to 11.7 at the same time.
Our previous paper reported the performance tests of the silica-removal from the geothermal brine by the silica treatment reactor. In this study conventional tracer tests were conducted in the reactor in order to obtain mathematical models that explain the transport of the geothermal brine through the reactor. An impulse of NaCl tracer was injected into the reactor together with geothermal brine. The concentration of the tracer was measured as an electrical conductivity at two points in the reactor. The numerical responses calculated by the mathematical model were fitted to the experimental responses. The model assumed the reactor to consist of six domains. One of three types of fluid flow conditions is considered in each domain. These three types of the fluid flowing were the perfect mixing flow, flow with dispersion, and ideal plug-flow. Numerical responses obtained by the model show good agreement with the experimental responses in the reactor. Reaction rate equations were appended to the model in order to estimate the results of the silica-calcium reaction in the reactor. The reaction rate equations and the reaction rate values were made from results in a batch reactor. The model with the reaction rate equation was successfully estimated the results of silica-calcium reaction in the reactor.
AE events accompanying hydraulic injection experiments at the Hijiori hot dry rock site was monitored by a network of ten borehole stations deployed at an average distance of 2 km from an injection well. AE events induced during the 1988 and 1992 hydraulic fracturing experiments with a high injection pressure were located near the injection point in the early stage of the experiments and clearly migrated towards the east and distributed along a vertical plane. The strike of epicenter distribution of AE vents is nearly parallel to the direction of the maximum principal stress. AE events induced during the 1989 and 1995 circulation tests with a low injection pressure were diffused. The permeability was estimated from the hypocenter migration as 10-16m2, which is intermediate between the permeability of core samples of granodiorite taken from a production well and the permeability of fractured rocks obtained by an injection test between the injection well and the production well. This indicates that the diffusion of AE events accompanying the circulation test is due to the permeation of water into joints which slip when the effective stress is reduced by the increased pore fluid pressure accompanying the hydraulic injection. The stress state at Hijiori was estimated based on the inversion from focal mechanisms of AE events. The best fit stress model obtained by inverting 58 focal mechanisms of AE events simultaneously indicates that the maximum principal stress σ1 is vertical, while the minimum principal stress σ3 is horizontal and trends north-south. The stress estimates obtained by the focal mechanism inversion essentially agree with other stress estimates previously obtained. It is therefore concluded that the focal mechanism inversion method provides a useful tool for estimating the stress state. Source parameters of AE events associated with the 1995 experiment were analyzed. Seismic moments and source radii estimated from the spectra of AE waveforms range from 8 to 585×107 N⋅m and 8 to 36 m, respectively. The S-wave Q is estimated to be 191. Comparison of the present source parameters with those reported for a similar magnitude range in a hard-rock formation indicates that our estimates of seismic moment and corner frequency are comparable.
D/SC (Deterministic and Stochastic Crack network simulator) has been developed with an objective to make reliable fracture network models by integrating various kinds of fracture information such as core analysis, borehole measurements, microseismic observations and geological surveys. The basic concept of D/SC is to put fractures into the model directly if their features have been determined and to interpolate the space in between by stochastic fractures, which follow the statistical properties of the reservoir. The model with a number of discrete fractures is then converted to a continuum model whose elements have equivalent permeability to the discrete fracture network. Finally, the pressure, flow vector and temperature at every element is calculated. The advantage of this approach is that the various kinds of data obtained from the reservoir can be utilized in describing the characteristics of the reservoir and that the created model can be used for predictions not only for reservoir performances in time but also for special potentials for new production wells. This paper describes a procedure of creating a fracture network model by applying D/SC to the Hijiori Hot Dry Rock reservoir. The Hijiori fracture model created includes several deterministic fractures, which are inferred from PTS logging data and BHTV images, in filling stochastic fractures, that follow statistical information derived from core analysis and BHTV images, and two permeable faults that represent both the caldera faults and the boundary of the microseismic distribution. The simulated thermal performance shows a significant difference in temperature draw down of production fluid from the two production wells (HDR-2a and HDR-3). The model was also used for the evaluation of pseudo-new wells in the Hijiori site. The results provided us with a guideline of preferable locations for an additional well.
The geologic structure in Hijiori field and the fracture characteristics of HDR reservoir were revealed by the areal geologic survey an well logging, which contained the information of orientedcores, BHTV images and PTS logs. There are seven geologic units from A to G in this area. Uppermost unit A is related to Holocene Hijiori Volcano that erupted about 10, 000 years ago with the caldera collapse. Unit G is the geologic basement of Cretaceous granodiorite whose fracture system is the base of the reservoir. In the HDR test site, near southern rim of the Hijiori caldera, top of the basement lies at 1, 445m through 1, 480m in vertical depth. Fractures striking N50°W to N90°W with steep dipping angle more than 70°dominate in the basement granodiorite. The predominant fractures in the granodiorite are main fluid flowing paths of the injected circulating water. The growth direction of the HDR reservoir is strongly controlled by the distribution of favorably oriented pre-existing fractures and their interaction with the stress field.