Blended cements containing mixtures of granulated blast-furnace slag (BFS) and Portland cement give low permeability matrices with initially favourable leach characteristics. Their retention arises from a combination of physical and chemical effects which include high pH and sorption. It is not practical though, in either laboratory or site-based experimentation, to determine changes in matrix chemistry over long timescales and the development of realistic models for long-term property predictions therefore becomes increasingly important. This paper pursues the development of such a model enabling changes in chemical and mineralogical balances during ageing to be predicted.

Phase development is assessed in the system CaO-Al2O3-SiO2-MgO-H2O. In the relevant composition range, the phases occurring include crystalline hydrates: portlandite, (Ca(OH)2); gehlenite hydrate, (2CaO.Al2O3.SiO2.8H2O); a hydrotalcite-structured phase (nominally 6MgO.Al2O3.(OH)x.yH2O). an AFm type phase, (nominally 4CaO.Al2O3 SO3.12H2O); and a poorly crysiallised c~lcium silicate hydrogel, C-S-H. All five phases are observed to occur together in slag-cements which are still hydrating. Given, as input, the chemical analyses of both the slag and the cement, and the initial blending proportions, the model predicts the equilibrium distribution between the five components and additionally, the Ca/Si ratio of the C-S-H. The aqueous chemistry in the system is predicted from the calculated phase distribution and appropriate solubility products.

Recent work by the authors has demonstrated that cesium was being incorporated in a hydrated phase as their cement-based waste forms cured. The objective of the present study was to identify the hydrated phases responsible for the observed cesium incorporation. Four calcium aluminosilicate glasses having a 1:2 Al2 O3 /CaO ratio and a 25–70 mole % silica content were mixed with ˜0.3 molar CsOH or CsCl solutions (water/solid ratio = 10) and allowed to hydrate at 38° and 90°C for periods up to 90 days. Paste equivalent samples, having a water/solid ratio of 1.0 were also prepared to gauge the cementing properties of these mixtures. Solutions were analyzed for Ca, Al, Si, and Cs while solids were characterized using x-ray powder diffraction and scanning electron microscopy.

Glasses exposed to the CsOH solution were more reactive than their counterparts exposed to the CsCl solution. In addition, reactivity as well as crystallinity seemed to be higher at 90° than at 38°C. At least two cesium-containing zeolites were identified in the 90°C CsOH solution experiment: A cesium-containing wairarkite analogue and possibly another cesium-containing zeolite unidentified at this time. The relatively complete removal of cesium from the solution, in one case, as well as the fact that the mixtures are self-cementing suggests that the glasses may be “engineered” to serve as overpack material in a deep-seated geologic repository.

The leachabilities of technetium and nitrate wastes immobilized in cement-based materials (i.e., grouts) have been investigated using ANS 16.1 test procedures. Factors found to affect the leachabilities include (1) grout mix ratio, (2) grout fluid density, (3) dry solid blend composition (including ground blast furnace slag), and (4) waste concentration.

This paper reports the results of some recent experimental studies of the solubility and sorption behaviour of lead-210 and carbon-14 under cementitious near-field conditions.

These studies have shown that under these conditions carbon-14 will have a maximum solubility limit of 10−4 M and that the distribution ratio, RD, will increase with increasing carbon-14 concentrations from 10−9 to 10−7 M. Not all of the carbon in the cement is available for exchange with carbon in the pore water. Differences in values of RD are observed between the two cement grout types studied, SRPC and OPC/BFS. Lead has been shown to have a maximum solubility limit of about 10−3 M at high pH. Good agreement is obtained between these measurements and thermodynamic modelling using the PHREEQE code. No differences were observed between lead solubilities under reducing or oxidising conditions at high pH values using the same phase separation techniques. Lead is particularly sensitive to the phase separation techniques employed. A factor of up to 250 difference is observed between 25000 and 30000 molecular weight cut-off filters. The values of RD for lead increase with decreasing lead concentrations and the values of RD for 10−3 M solutions are observed to be 500 mlg−1 for SRPC and 1300 mlg−1 for OPC/BFS.

The effect of a concrete-based repository on local geological conditions and properties has been explored. As part of a complete description of repository behaviour, a static model of clay-alkali interactions has been set up. Parallel experiments have demonstrated the buffering capacity of clays, and are indicating the chemical and physical changes that may occur when an alkali plume interacts with a clay host rock.

Candidate products are being evaluated to immobilize the routinely calcined waste at the Idaho Chemical Processing Plant (ICPP). A potential product with minimal volume for immobilizing ICPP high-level waste (HLW) for final disposal is a high-waste-loading and high-density glass-ceramic. Glass-ceramics are formed by Hot Isostatic Pressing (HIPing) the HLW with selected additives, such as SiO2, B2O3, Li2O, Na2O, and Y2O3. Glass-ceramic products have been formed with calcine loa ings up to 80 wt% and densities up to 3.4 g/cm3. Crystalline phases observed in the glass-ceramic products include calcium fluoride, monoclinic and cubic zirconia, calcium- and yttrium-stabilized zirconia, and zircon. An interstitial amorphous phase also exists consisting of the oxides of silicon, aluminum, boron, and alkalis. The glass-ceramic waste forms give leach rates comparable to simulated HLW glass products.

Approximately 400 × 106 liters of low-level alkaline salt solution will be treated at the Savannah River Plant (SRP) Defense Waste Processing Facility (DWPF) prior to disposal in concrete vaults at SRP. Treatment involves removal of Ca+and Sr+2 followed by solidification and stabilization of potential contaminants in saltstone, a hydrated ceramic waste form.

Chromium, technetium, and nitrate releases from saltstone can be significantly reduced by substituting hydraulic blast furnace slag for portland cement in the formulation designs. Slag-based mixes are also compatible with Class F fly ash used in saltstone as a functional extender to control heat of hydration and reduce permeability. (Class F fly ash is locally available at SRP.)

A monolithic waste form is produced by the hydration of the slag and fly ash. Soluble ion release (NO3−) is controlled by the saltstone microstructure (bulk porosity and pore size and connectivity). Chromium and technetium are 103−104 times less leachable from slag mixes than from cement-based waste forms. Results suggest that chemical stabilization rather than physical entrapment, as in the case of nitrate is responsible for this improved leaching. Reduction of Cr+6 and Tc+7 to Cr+3 and Tc+4 by ferrous iron or Mn+2 in the slag and subsequent precipitation of the relatively insoluble phses Cr(OH)3 and TcO2 are proposed as the stabilizing reactions.

A review of chemical phenomena of importance for the conceptual near-field modelling of nuclear waste repositories is given. The emphasis is on thermodynamic modelling and the coupling of chemical processes mediated through ground water.

The prediction of the long term corrosion of metal containers for nuclear waste under geological disposal conditions requires the extrapolation of corrosion data over several hundred years. There is a general trend to use either simulation or ‘worst case’ experiments as a means of materials selection, but neither really addresses the problem of long term extrapolation. It is proposed that such an extrapolation can only be done convincingly if it is based on a sound and generally accepted mechanistic understanding of the processes involved. If such knowledge does not exist the first step must be to acquire it through experimental mechanistic studies. Subsequently such knowledge should be formulated into mathematical models, which can be used to make long term predictions, and which can be validated by comparison with short term experimental results. The application of this combined mathematical modelling/experimental approach is illustrated for three corrosion processes which may affect carbon steel containers, namely general corrosion, localised attack and stress corrosion cracking.

The concept of disposal of unreprocessed spent fuel has now been under study internationally for over ten years. Considerable progress has been made in understanding the factors that will control radionuclide release from spent fuel in an underground disposal vault. This progress is reviewed and the research areas of significance in providing further data for source term models are discussed. Key areas for future research are identified; these include improved characterization of spent fuel to determine the inventories of fission products at grain boundaries, together with their release kinetics; and a better understanding of the effects of solution chemistry on spent fuel dissolution, in particular the effects of salinity, redox chemistry, and radiolysis of groundwater. Approaches to modelling the dissolution of spent fuel are discussed, and a possible approach for developing an oxidative dissolution model is outlined.

For a number of years, nuclear regulators have grappled with difficult questions such as: “How safe is safe enough?” Such issues take on new dimensions in the long time-frame of high-level waste disposal.

Many of the challenges facing regulators involve assessment of long-term materials performance. Because real-time experiments cannot be conducted, it is necessary to rely extensively on modeling. This raises issues regarding the extent to which long-term extrapolations of short-term data are justified, the question of how closely models must represent reality to be trusted, and practical matters such as methods for validating unique computer codes.

Issues such as these illustrate how regulators must make decisions in a climate of uncertainty. Methods used by non-technical disciplines to make decisions under uncertainty have been examined and offer solutions for regulators and licensees alike.

A primary purpose of performance assessment of geologic repositories for radioactive waste is to predict the extent to which radioactive species are released from the waste solids and are transported through geologic media to the environment. Reliable quantitative predictions must be made of rates of release of radionuclides from the waste into the rock, transport through the geologic media, cumulative release to the accessible environment, and maximum concentrations in ground water and surface water. Here we review theoretical approaches to making the predictions of near-field release from buried waste solids, which provide the source terms for far-field release. The extent to which approaches and issues depend on the rock media and on regulatory criteria is discussed.

Results are presented from a migration experiment carried out in a natural fracture in a 1 m × 1 m × 0.6 m quarried granite block. Near-uniform flow in the fracture was achieved by controlling the groundwater flow from the fracture laterally across the exit face of the fracture. The longitudinal dispersion was determined from the elution profile of uranine, a non-sorbing tracer. The effect of matrix diffusion was determined by reducing the linear velocity of the transport solution from approximately 3 to 0.75 cm/h. The velocity of a sorbing tracer, Cs-137, was compared to that of non-sorbing I-131 and with that calculated from data obtained for Cs-137 in static sorption experiments on similar fracture surfaces. Information on channeling in the fracture was obtained by autoradiographing the fracture surface of the separated block after the termination of the experiment. Quantitative information on the sorbed Cs-137 inventory by two-dimensional gamma scanning of the fracture surfaces is underway.

Radionuclide transport through fractured media is usually calculated assuming that water flows in most of the fractures. Several observations in the field and the laboratory show that flow is very unevenly distributed in fractured crystalline rock. These observations indicate that most of the water flow takes place in a limited number of channels. The channels are seldom wider than a few meters and are often much narrower. This means that the surface of the fracture in contact with the flowing water (wetted surface) is less than one might expect.

This low value of the wet surface of the fracture may considerably influence the transport of radionuclides through fractured media. If the channels do not intersect over a certain distance, then the channels may be modelled as a bundle of independent channels. Channels with a large flow and small sorption surface will carry the tracer rapidly and in large amounts.

Calculations are performed for cases where channeling is assumed to take place. The most important entities to assess are the water flow distribution in the different channels, the wetted surface of the channels, the diffusivity into the rock matrix, and the sorption coefficient in the matrix. Experimental data for the water flow distribution are used and the transport of nuclides is calculated for the different channels. From these values the concentration of the effluent is determined. The results show that the retardation for the nonsorbing nuclides is negligible. Retardation is only important for the nuclides which are strongly sorbed on the granitic rock. Calculations are also done assuming other channel frequencies and other overall water flowrates.

This paper describes the programme for an in-situ radionuclide migration experiment in fractured crystalline rock and presents the first results of site characterisation work. This study is particularly notable for its extensive hydrologic/geochemical support programme and the use of radiotracers identified as important in safety assessment studies.

Three similar migration experiments in the matrix of granitic rock are presented. The experiments have been carried out in “undisturbed” rock, that is rock under its natural stress environment. Since the experiments were performed at the 360 m level (in the Stripa Mine), the rock was subjected to nearly the same conditions as the rock surrounding a nuclear waste repository as proposed in the Swedish concept (SKB).

A mixture of three non-sorbing (conservative) tracers, Uranine, Cr-EDTA and I−, were injected into the granitic rock matrix for time periods of months up to years. The subsequent overcorings of the injection holes showed that the tracers had in some cases migrated at least ≈ 400 mm (measuring limit) into the rock matrix for the experiment with the longest injection time. It could also be seen that there were large differences in migration distance into the rock matrix for samples taken fairly close to each other. One example where the tracers have diffused through fissure coating (filling) material located in “undisturbed” rock is also presented.

The results from all three experiments show that all three tracers have migrated through the disturbed zone close to the injection hole, through the fissure coating material and a distance into the “undisturbed” rock matrix.

These results therefore indicate that dissolved compounds may migrate into the rock matrix. This migration into the rock matrix will increase the area available for sorption of radionuclides significantly and may therefore increase the migration times for radionuclides by order(s) of magnitude.

Diffusivities and hydraulic conductivities obtained in this in-situ experiment compare well with those obtained in laboratory experiments.

A large scale tracer experiment has been performed in sparsely fractured granitic rock at 360 m depth. Nine different conservative tracers were injected at distances ranging between 10 and 56 m from a drift excavated for this experiment. The upper part of the test site was covered completely with plastic sheets. All the water entering into the drift could be collected. Water flow monitoring shows that water flow takes place in small areas with large dry areas in between. Of the nine injected tracers six reached the test site during the experiment. The fitting of the breakthrough curves to different models could not satisfactorily explain the actual processes involved in the tracer migration. The results of the tracer experiments and the Tritium measurements gave strong support to the notion that a non–negligible portion of the flow takes place in more or less isolated channels.

Brine flow has been measured to unheated boreholes for periods of a few days and to heated holes for two years in the WIPP facility. It is proposed that Darcy flow may dominate the observed influx of brine. Exact solutions to a linearized model for one-dimensional, radial flow are evaluated for conditions approximating the field experiments. Flow rates of the correct order of magnitude are calculated for permeabilities in the range 10−21–1020 m2 (1–10 nanodarcy) for both the unheated and heated cases.

Under the auspices of the OECD/NEA Stripa Project, a state-of-the-art evaluation of repository sealing materials and techniques has been undertaken with particular application to crystalline rock. The objectives of the evaluation were (1) to review progress to date in the development of repository sealing materials and emplacement techniques; (2) to identify priority materials for sealing narrow aperture fractures in crystalline rock, including issues of technical concern to the long-term effectiveness of such materials; and (3) to provide a framework for advancing an in situ fracture-sealing test program that can be designed and implemented at the Stripa Mine.

The classes of sealing materials that were considered included cementitious materials, natural clay materials, chemical grouts, fracture-filling synthetic minerals, ceramics, and metals. Of these, cementitious materials and natural clay materials were recommended as high-priority materials for further study because they can be designed to meet desired repository performance characteristics, there is considerable history of successful use in similar engineering applications, and there is indirect evidence that they will continue to perform as expected for long periods of time. Techniques for the emplacement of these sealing materials in rock fractures include pressure injection, dynamic (or vibratory) injection, and electrophoresis.

Defense-generated transuranic (TRU) waste will be stored at WIPP in the bedded halite of the Salado Formation (Permian), which is overlain by the impure Permian evaporites of the Rustler Formation and the Dewey Lake Red Beds. Both the Rustler and Dewey Lake contain abundant to less common secondary selenite veins of uncertain origin, and dissolution zones occur in the Rustler. The Rustler Formation also contains two dolomite aquifers, the Magenta and Culebra members. The purpose of this study is to determine whether vein selenite is locally derived, or has been introduced in moving groundwater solutions. We have used Sr isotopic studies and REE, U and other trace element data to address the problem. The Sr isotopic data show that neither the Salado nor Rustler anhydrites have exchanged with secondary sources of Sr, and this is supported by the REE and U data. Further, selenite veins from the Rustler possess Sr isotopic compositions identical to the Rustler, indicating local origin, and this is also supported by the U and REE data. Selenite veins from the Dewey Lake Red Beds possess Sr isotopic compositions closer to surface caliche deposits, and may contain near-surface derived Sr. U and REE data show more scatter and evidence for extrinsic sources than the vein/host-rock pairs from the Rustler. Sr isotopic compositions of the Magenta and Culebra indicate that the Magenta has undergone less water/rock interaction than the Culebra. Collectively, the data argue for more possible surface or near-surface water/rock interactions in the Dewey Lake Red Beds than in the Rustler Formation; such interactions in the Salado Formation have been minimal.