Deep geological disposal is a widely accepted approach for safe management and long-term disposal of high-level radioactive waste (HLW). However, high uncertainty associated with subsurface properties of fractured rocks is a significant obstacle to practical safety assessment of HLW disposal. In this study, we develop an integrated statistical framework for uncertainty quantification of radionuclide migration related to the geological disposal of HLW. We employ a response surface methodology integrated with Monte Carlo simulations of radionuclide migration in fractured granite to perform global sensitivity and statistical analysis by coupling the uncertainty quantification tool, PSUADE, and radionuclide migration simulator, FRACPIPE. FRACPIPE is a semi-analytical simulator that models solute transport in fractures and matrix slabs with an arbitrary-length decay chain and an arbitrary time-varying influent concentration history by considering a variety of transport mechanisms. The statistical risk metrics include nuclide breakthrough (BT) time, total release dose, single nuclide release dose, and single nuclide flux. The global sensitivity analysis identifies fracture aperture, matrix diffusion coefficient, hydraulic gradient, and dispersivity as the most sensitive parameters. Considering the uncertainty ranges for independent variables (e.g., dispersivity, hydraulic gradient, fracture spacing, fracture aperture, sorption distribution coefficient, matrix diffusion, and matrix porosity), we apply post-processing results of Monte Carlo simulations to conduct statistical analysis of the risk metrics. Under this studied condition, the average BT time for radionuclides at 1000 m is between 85 and 331 thousand years. This study provides valuable insights into the impact of input parameter (independent variables) uncertainty and sensitivity on radionuclide migration behavior in an HLW repository.
Abstract Groundwater monitoring networks are direct sources of information for revealing subsurface system dynamic processes. However, designing such networks is difficult due to uncertainties in the spatial heterogeneity of aquifer parameters such as permeability ( k ). This study combines deep learning and information theory with an optimization framework to address network design problems in heterogeneous aquifer systems. The framework first employs a generative adversarial network to parameterize heterogeneous k distribution using a low‐dimensional latent representation. Then, surrogate models are developed based on the deep neural networks to perform uncertainty quantification of pressure heads and solute concentrations at locations of pre‐designed candidate monitoring stations. The monitoring stations are then ranked using the greedy search algorithm based on the maximum information minimum redundancy (MIMR) criterion. In order to depict the importance of each candidate monitoring location, the hotspot maps of the selection probability ( P s ) are derived from MIMR repetition results. Comprehensive monitoring networks derived from the hotspot maps are then conducted as the final monitoring stations to improve monitoring information compared to MIMR results. Additionally, nine entropy quantization strategies are compared to evaluate their effects on monitoring network optimization results. Results indicate that caution should be taken when selecting entropy quantization strategies to achieve the accuracy required for model calibration and to improve the efficiency of monitoring optimization. Considering high‐dimensional uncertainties associated with aquifer parameters, the developed framework can provide important insights for monitoring network designs in various earth observational projects.
Promoting sustainable mining practices while safe-guarding water ecosystems demands precise anticipation of mine water influx. This investigation pioneers a novel approach harnessing microseismic monitoring to detect water-conducting conduits and elevate proactive response strategies. Through the utilization of microseismic energy density analysis, fracture points within rock formations are continuously monitored, offering real-time insights. Nonetheless, the data generated from this method often exhibits fragmentation, sporadic patterns, and data heterogeneity, complicating the identification of evolving water-conducting pathways. To surmount this challenge, we have seamlessly integrated the Self-Attention mechanism into the Long Short-Term Memory (LSTM) model, resulting in the innovative SA-LSTM fusion. This hybrid model predicts the following day's water inflow, effectively merging data from microseismic monitoring with groundwater levels. This fusion facilitates a robust correlation between monitoring data and water inflow metrics. Comparative assessments underscore the SA-LSTM's superiority over other intricate time-series models in terms of forecast precision, with a MAE of 21.8 m 3 /h, RMSE of 39.3 m 3 /h and MAPE of 2.8% in the test stage of the water inflow event. By amalgamating diverse datasets, it substantially enhances the accuracy of predicting water inflow within coal mines. The discernments from this study not only introduce more accurate water inflow predictions but also provide technical guidance for the safety production of mine.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)
