Explore the vast range of titles available.

Carbonation of alkali activated materials is one of the main deteriorations affecting their durability. However, current understanding of the structural alteration of these materials exposed to an environment inducing carbonation at the nano/micro scale remains limited. This study examined the evolution of phase assemblages of alkali activated slag mortars subjected to accelerated carbonation (1% CO 2 , 60% relative humidity, up to 28 day carbonation) using XRD, FTIR and 29 Si, 27 Al, and 23 Na MAS NMR. Samples with three water to binder (w/b) ratios (0.35, 0.45, and 0.55) were investigated. The results show that the phase assemblages mainly consisted of C-A-S-H, a disordered remnant aluminosilicate binder, and a minor hydrotalcite as a secondary product. Upon carbonation, calcium carbonate is mainly formed as the vaterite polymorph, while no sodium carbonate is found after carbonation as commonly reported. Sodium acts primarily as a charge balancing ion without producing sodium carbonate as a final carbonation product in the 28-day carbonated materials. The C-A-S-H structure becomes more cross-linked due to the decalcification of this phase as evidenced by the appearance of Q 4 groups, which replace the Q 1 and Q 2 groups as observed in the 29 Si MAS NMR spectra, and the dominance of Al(IV) in 27 Al MAS NMR. Especially, unlike cementitious materials, the influence of w/b ratio on the crystalline phase formation and structure of C-A-S-H in the alkali activated mortars before and after carbonation is limited.

We accelerate reactive transport (RT) simulation by replacing the geochemical solver in the RT code by a surrogate model or emulator, considering either a trained deep neural network (DNN) or a k -nearest neighbor (kNN) regressor. We focus on 2D leaching of hardened cement paste under diffusive or advective-dispersive transport conditions, a solid solution representation of the calcium silicate hydrates and either 4 or 7 chemical components, and use the HPx (coupled Hydrus-PHREEQC model) reactive transport code as baseline. We find that after training, both our DNN-based and kNN-based codes, HPx py -DNN and HPx py -kNN, can make satisfactorily to very accurate predictions while providing either a 3 to 9 speedup factor compared to HPx with parallelized geochemical calculations over 4 cores. Benchmarking against single-threaded HPx, these speedup factors become 8 to 33. Overall, HPx py -DNN and HPx py -kNN are found to achieve a close to optimal speedup when DNN regression and kNN search are performed on a GPU. Importantly, for the more complex 7-components cement system, no emulator that is globally accurate over the full space of possible geochemical conditions could be devised. Instead we therefore build “local” emulators that are only valid over a relevant fraction of the input parameter space. This space is identified by running a coarse and thus computationally cheap full RT simulation, and subsequently explored by kernel density sampling. Future work will focus on improving accuracy for this type of cement systems.

HPx is a multicomponent reactive transport model which uses HYDRUS as the flow and transport solver and PHREEQC-3 as the biogeochemical solver. Some recent adaptations have significantly increased the flexibility of the software for different environmental and engineering applications. This paper gives an overview of the most significant changes of HPx, such as coupling transport properties to geochemical state variables, gas diffusion, and transport in two and three dimensions. OpenMP allows for parallel computing using shared memory. Enhancements for scripting may eventually simplify input definitions and create possibilities for defining templates for generic (sub)problems. We included a discussion of root solute uptake and colloid-affected solute transport to show that most or all of the comprehensive features of HYDRUS can be extended with geochemical information. Finally, an example is used to demonstrate how HPx, and similar reactive transport models, can be helpful in implementing different factors relevant for soil organic matter dynamics in soils. HPx offers a unique framework to couple spatial-temporal variations in water contents, temperatures, and water fluxes, with dissolved organic matter and CO[2] transport, as well as bioturbation processes.

This paper presents a detailed comparison between 3 methods for emulating CPU-intensive reactive transport models (RTMs): Gaussian processes (GPs), polynomial chaos expansion (PCE), and deep neural networks (DNNs). State-of-the-art open source libraries are used for each emulation method while the CPU-time incurred by one forward run of the considered RTMs varies from 1 h to between 1 h and 30 min and 5 days. Besides direct emulation of the simulated uranium concentration time series, replacing the original RTM by its emulator is also investigated for global sensitivity analysis (GSA), uncertainty propagation, and probabilistic calibration using Markov chain Monte Carlo (MCMC) sampling. The selected DNN is found to be superior to both GPs and PCE in reproducing the input–output behavior of the considered 8-dimensional and 13-dimensional CPU-intensive RTMs. This even though the used training sets are small, from 75 to 500 samples. Furthermore, the two used PCE variants, standard PCE and sparse PCE (sPCE), appear to always provide the least accuracy while not differing much in performance. As a consequence of its better emulation capabilities, the DNN method outperforms the two other methods for uncertainty propagation. For the GSA application, the DNN and GP methods offer equally good approximations to the true first-order and total-order Sobol’ sensitivity indices while PCE does less well. Most surprisingly, despite its superior emulation skills, the DNN approach leads to the worst solution of the considered synthetic inverse problem which involves 1224 measurement data with low noise. This apparently contradicting behavior of the used DNN is at least partially due to the small but complicated deterministic noise that affects the DNN-based predictions. Indeed, this complex error structure can drive the emulated solutions far away from the true posterior distribution in case of high-quality measurement data. Among the considered 3 methods, only the GP method allows for retrieving emulated posterior solutions that jointly (1) fit the high-quality measurement data to the appropriate noise level (log-likelihood value) and (2) most closely fit the true model parameter values. Overall, our findings indicate that when the available training set is relatively small (75 to 500 input-output examples) and fixed beforehand, PCE is not the best choice for emulating CPU-intensive RTMs. Instead, GPs or DNNs should be preferred. However, a DNN can deliver overly biased model calibration results. In contrast, the GP method performs fairly well across all considered tasks: direct emulation, global sensitivity analysis, uncertainty propagation, and calibration.

Components of the so-called “multiple-barrier system” from the waste form to the biosphere include a combination of waste containers, engineered barriers, and natural barriers. The Engineered Barrier System (EBS) is crucial for containment and isolation in a radioactive waste disposal system. The number, types, and assigned safety functions of the various engineered barriers depend on the chosen repository concept, the waste form, the radionuclides waste inventory, the selected host rock, and the hydrogeological and geochemical settings of the repository site, among others. EBS properties will evolve with time in response to the thermal, hydraulic, mechanical, radiological, and chemical gradients and interactions between the various constituents of the barriers and the host rock. Therefore, assessing how these properties evolve over long time frames is highly relevant for evaluating the performance of a repository system and safety function evaluations in a safety case. For this purpose, mechanistic numerical models are increasingly used. Such models provide an excellent way for integrating into a coherent framework a scientific understanding of coupled processes and their consequences on different properties of the materials in the EBS. Their development and validation are supported by R&D actions at the European level. For example, within the HORIZON 2020 project BEACON (Bentonite mechanical evolution), the development, test, and validation of numerical models against experimental results have been carried out in order to predict the evolution of the hydromechanical properties of bentonite during the saturation process. Also, in relation to the coupling with mechanics, WP16 MAGIC (chemo Mechanical AGIng of Cementitious materials) of the EURAD Joint Programming Initiative focuses on multi-scale chemo-mechanical modeling of cementitious-based materials that evolve under chemical perturbation. Integration of chemical evolution in models of varying complexity is a major issue tackled in the WP2 ACED (Assessment of Chemical Evolution of ILW and HLW Disposal cells) of EURAD. WP4 DONUT (Development and improvement of numerical methods and tools for modeling coupled processes) of EURAD aims at developing and improving numerical models and tools to integrate more complexity and coupling between processes. The combined progress of those projects at a pan-European level definitively improves the understanding of and the capabilities for assessing the long-term evolution of engineered barrier systems.

The paper presents an evaluation of the combined use of the HYDRUS and SWI2 packages for MODFLOW as a potential tool for modeling recharge in coastal aquifers subject to saltwater intrusion. The HYDRUS package for MODFLOW solves numerically the one-dimensional form of the Richards equation describing water flow in variablysaturated media. The code computes groundwater recharge to or capillary rise from the groundwater table while considering weather, vegetation, and soil hydraulic property data. The SWI2 package represents in a simplified way variable-density flow associated with saltwater intrusion in coastal aquifers. Combining these two packages within the MODFLOW framework provides a more accurate description of vadose zone processes in subsurface systems with shallow aquifers, which strongly depend upon infiltration. The two packages were applied to a two-dimensional problem of recharge of a freshwater lens in a sandy peninsula, which is a typical geomorphologic form along the Baltic and the North Sea coasts, among other places. Results highlighted the sensitivity of calculated recharge rates to the temporal resolution of weather data. Using daily values of precipitation and potential evapotranspiration produced average recharge rates more than 20% larger than those obtained with weekly or monthly averaged weather data, leading to different trends in the evolution of freshwater-saltwater interfaces. Root water uptake significantly influenced both the recharge rate and the position of the freshwater-saltwater interface. The results were less sensitive to changes in soil hydraulic parameters, which in our study were found to affect average yearly recharge rates by up to 13%.

The use of the subsurface for CO 2 storage, geothermal energy generation, and nuclear waste disposal will greatly increase the interaction between clay(stone) and concrete. The development of models describing the mineralogical transformations at this interface is complicated, because contrasting geochemical conditions (Eh, pH, solution composition, etc.) induce steep concentration gradients and a high mineral reactivity. Due to the complexity of the problem, analytical solutions are not available to verify code accuracy, rendering code intercomparisons as the most efficient method for assessing code capabilities and for building confidence in the used model. A benchmark problem was established for tackling this issue. We summarize three scenarios with increasing geochemical complexity in this paper. The processes considered in the simulations are diffusion-controlled transport in saturated media under isothermal conditions, cation exchange reactions, and both local equilibrium and kinetically controlled mineral dissolution-precipitation reactions. No update of the pore diffusion coefficient as a function of porosity changes was considered. Seven international teams participated in this benchmarking exercise. The reactive transport codes used (TOUGHREACT, PHREEQC, with two different ways of handling transport, CRUNCH, HYTEC, ORCHESTRA, MIN3P-THCm) gave very similar patterns in terms of predicted solute concentrations and mineral distributions. Some differences linked to the considered activity models were observed, but they do not bias the general system evolution. The benchmarking exercise thus demonstrates that a reactive transport modelling specification for long-term performance assessment can be consistently addressed by multiple simulators.

The HYDRUS-1D and HYDRUS (2D/3D) computer software packages are widely used finite element models for simulating the one-, and two- or three-dimensional movement of water, heat, and multiple solutes in variably-saturated media, respectively. While the standard HYDRUS models consider only the fate and transport of individual solutes or solutes subject to first-order degradation reactions, several specialized HYDRUS add-on modules can simulate far more complex biogeochemical processes. The objective of this paper is to provide a brief overview of the HYDRUS models and their add-on modules, and to demonstrate possible applications of the software to the subsurface fate and transport of chemicals involved in coal seam gas extraction and water management operations. One application uses the standard HYDRUS model to evaluate the natural soil attenuation potential of hydraulic fracturing chemicals and their transformation products in case of an accidental release. By coupling the processes of retardation, first-order degradation and convective-dispersive transport of the biocide bronopol and its degradation products, we demonstrated how natural attenuation reduces initial concentrations by more than a factor of hundred in the top 5 cm of the soil. A second application uses the UnsatChem module to explore the possible use of coal seam gas produced water for sustainable irrigation. Simulations with different irrigation waters (untreated, amended with surface water, and reverse osmosis treated) provided detailed results regarding chemical indicators of soil and plant health, notably SAR, EC and sodium concentrations. A third application uses the HP1 module to analyze trace metal transport involving cation exchange and surface complexation sorption reactions in a soil leached with coal seam gas produced water following some accidental water release scenario. Results show that the main process responsible for trace metal migration in soil is complexation of naturally present trace metals with inorganic ligands such as (bi)carbonate that enter the soil upon infiltration with alkaline produced water. The examples were selected to show how users can tailor the required model complexity to specific needs, such as for rapid screening or risk assessments of various chemicals nder generic soil conditions, or for more detailed site-specific analyses of actual subsurface pollution problems.

The benchmark problem presented in this paper deals with the leaching of calcium from hardened cement paste. The leaching of calcium results in the dissolution of the cement minerals which affects physical, chemical and mechanical properties of porous cement matrix. The dissolution of cement minerals in this case progresses heterogeneously as a consequence of a small-scale geometrical feature (crack) within a domain. Complexity of transport through cracked porous media combined with complex cement chemistry can lead to considerable modelling uncertainties. One possible way to get an insight into the robustness of modelling results is to perform benchmark based on (i) different transport models and solution methods (finite volume, finite element, etc.), (ii) different geochemical solvers and (iii) different coupling algorithms (sequential iterative and non-iterative). This benchmark is designed to gradually increase the complexity of the problem and in this way recognize modelling elements that are the most sensitive in terms of modelling results, e.g. evolution of physical and chemical properties. Five international teams participated in this benchmark exercise. The reactive transport codes used (HYTEC, MIN3P, OGS-GEM, ORCHESTRA, COMSOL Multiphysics-iPHREEQC) give similar patterns in terms of predicted concentrations of elements and the mineralogy. The level of agreement depends on the problem complexity related mainly to the weighting and conservation properties of different numerical methods, to the coupling between transport and reactive solver and the agreement of thermodynamic database.

A deep geological repository is considered an appropriate option for disposal of radioactive waste containing long-lived radionuclides. Engineered barriers’ degradation and radionuclide transport strongly depend on the conditions in the repository. This work presents the assessment of the geochemical evolution in a radioactive long-lived intermediate-level waste disposal cell constructed in granite. The considered cell consists of cemented waste packages, cementitious backfill and several meters of host rock. Three abstracted reactive transport models of different complexity were developed: a 1D model and a 2D model considering transport by advection and diffusion and a 2D model with diffusive transfer only. The changes in the pH, the pore water composition and the materials’ mineralogical composition were observed. The modelling results indicate an increase in the pH in the disposal cell due to leaching of alkalis, which is followed by the dissolution of portlandite and the precipitation of calcite at the granite-vault backfill boundary. The obtained changes in the pH indicate that the geochemical alterations in the disposal cell proceed very slowly. Such a slow degradation could be the result of the formation of the higher pH zone upstream from the disposal tunnel. The advection–diffusion models in 1D and 2D geometries produced similar results. However, the 2D model also identified spatial peculiarities in the changes of the geochemical environment. Comparison of the pure diffusive case results with the advection–diffusion cases demonstrated that both processes are relevant in the analysed disposal cell.