Abstract
Nagra is currently developing design concepts for canisters for the deep geological disposal of spent fuel and high-level waste. A feasibility evaluation study was done to assess a number of candidate canister designs and materials. The potential canister options were assessed with regard to long-term safety by considering mechanical integrity, environmental damage and potential impact on the geological barrier. Manufacturing feasibility, sealing, inspection as well as potential cost were also assessed. A canister concept, based on a thick-walled forged carbon steel substrate coated with copper, was favourably assessed. As a result, development work was undertaken to refine the copper-coated canister design. Different lid designs and closure configurations were considered, using finite element analysis, to determine the limit load by plastic yielding and buckling and to determine safety margins under the design load. In addition, finite element fracture analyses were performed to determine stress intensity factors at the closure weld root. A hemispherical lid with a partially penetrating weld at a depth of 25-32 mm, having a root gap in the axial direction, was found to be fit for purpose while offering the potential to avoid post-weld heat treatment. An evaluation of relevant welding processes for closure welding of the selected joint design was then carried out. The potential welding processes were ranked based on a number of criteria linked to the maturity of the technique, the quality of the resulting weld, the properties of the weld material, and the applicability of the technique to the current canister design and the requirements of deployment in a hot cell.
This paper is part of a supplement on the 6th International Workshop on Long-Term Prediction of Corrosion Damage in Nuclear Waste Systems.
Keywords
Introduction
Nagra's long-term plan for disposal of radioactive waste is to design and build repositories for spent fuel (SF) and high-level waste (HLW), and low- and intermediate-level waste (L/ILW), either separately or as a combined repository. At present, repository planning is proceeding through the Sectoral Plan for Geological Repositories, for which Stage 1 is complete and the host rock for the SF/HLW repository (Opalinus Clay) has been selected. By 2024, upon completion of the Sectoral Plan, Nagra intends to make General License applications for the two repositories. In the case of disposal of SF/HLW, repository construction is not expected to take place before 2050, with operation following around 2060 [1].
The SF/HLW canister option to be selected is an integral part of the multi-barrier system, thus it is important for the general license application to demonstrate the feasibility of materials choices and concepts, and provide evidence that operational and long-term safety can be ensured. Because the time to repository implementation is long, it is intended to take maximum advantage of developments elsewhere, including on-going developments in materials and technology. Thus, the selection of a final material and design is unlikely to be made until after the general license application. A wide range of designs and materials have been considered, such as a forged carbon steel canister (current reference), a carbon steel or cast iron canister coated with a corrosion-resistant material with a thickness of a few millimetres such as copper, nickel or titanium alloys, a copper canister similar to the KBS-3 design [2], as well as ceramic canisters.
The present paper summarises recent studies at Nagra, including the results of a broad assessment of materials and design options against generic canister requirements. This is followed by results of design, structural analysis and weld closure evaluation studies.
Disposal environment
The disposal canisters are expected to be emplaced at 600-900 m depth in horizontal tunnels constructed in the centre of a layer of Opalinus Clay, a low-permeability clay rock located in various potential siting regions in northern Switzerland. The canisters would be placed on bentonite blocks, with the void spaces around and between canisters backfilled with granular bentonite. After backfilling, the canister surface temperature would increase to a maximum of about 140°C and decrease gradually to the natural rock temperature after a few thousand years (see Figure 1).
Evolution of temperature, relative humidity and relative oxygen concentration at the canister surface with time for a repository in Opalinus Clay [3].
The pore-water pressure would gradually increase to hydrostatic in a few hundred years and the canisters could eventually experience the full lithostatic load (e.g. about 22 MPa if the repository were at a maximum depth of 900 m), with a potential for some anisotropy in the load. The pore-water is moderately saline (about 0.3 mol L−1 Cl) and the redox conditions are expected to rapidly become anoxic. More detail on the repository conditions and the corrosion behaviour of carbon steel and copper (the materials that have been studied most) is provided in Diomidis and Johnson [3].
Canister requirements
Based on the above-mentioned expected evolution of the near field conditions in the repository as well as operational safety needs, a set of requirements that a disposal canister needs to fulfil has been defined [3]. The key requirements are related principally to the long-term safety, the handling concept and manufacturing and include:
Minimum lifetime: There is a regulatory requirement of a minimum lifetime (no breach of containment) of 1000 years for SF and HLW canisters. However, Nagra proposes a lifetime requirement of 10 000 years in order to provide a significant margin of safety with respect to the regulatory target and to examine structural performance over a time frame significantly exceeding 1000 years. Wall thickness: The canister waste loading and shielding concept provide the basis for the shielding calculation and derived wall thickness. The wall thickness should ensure long-term structural integrity and that the radiation dose rate at the canister outer surface is <1000 mSv h−1 in order to preclude radiation-induced corrosion. Welding method: There must be a satisfactory method for welding or sealing applicable to a thickness in the range in question. The weld depth need not be equal to the entire wall thickness if the requirements related to the structural integrity of the weld are met. The welding method must be suitable for remote operation, given the radiation field. Gas production-corrosion rate: A requirement for the maximum rate of gas production in a SF/HLW repository in Opalinus Clay corresponds to a corrosion rate under anaerobic conditions which is <10 μm/year. This limit in gas production rate is set in order to avoid excessive pressure build-up in the repository which may compromise the integrity of the host rock. Structural integrity and inspectability: The stresses in the canister wall, lid and base, including the weld region should not give rise to structural failure causing breach of containment for at least 10 000 years as demonstrated by compliance with an appropriate design standard. The stresses in the weld region and heat-affected zone (HAZ) should be low enough to preclude the occurrence of stress-assisted failure processes such as stress corrosion cracking and hydrogen-induced cracking. The weld procedure should ensure any defects remaining in the canister after manufacturing are smaller than the critical crack length by a suitable margin, while the inspection process should be able to detect them. The post-weld inspection method must be suitable for remote operation, given the radiation field. The stress reduction method for a fully loaded and welded SF or HLW canister should not damage the SF or HLW. The guidelines used are SF temperature <400°C and HLW temperature <450°C. Sub-criticality: The loaded SF canisters should be subcritical when the internal void spaces are water-filled. Handling and retrievability: The canister must remain structurally sound without breach of containment during normal handling, retrieval and incidents that might occur during handling. Because the canister will have to be surrounded by a heavy thick-walled transfer/shielding overpack while in the encapsulation facility and during transfer underground, it is considered, in this condition, to have low vulnerability to impacts from handling. The stresses in the canister as a result of a handling incident should be less than values that would indicate a possibility of breaching determined on the basis of conservative assumptions and analyses. The canister must also satisfy the requirement for retrievability during the operational phase. This implies that it must be possible to grapple the canister and pull it out into a shielding unit. For this operation, it is assumed that the surrounding bentonite pellets have been removed, thus there is expected to be limited friction involved in pulling the canister.
For most of these, more detailed requirements and criteria for acceptability have been developed and these depend on the material under consideration. For example, in a design study for canisters for SF and HLW, Patel et al. [4] outlined such detailed requirements for the specific case of a design concept for thick-walled carbon steel.
Feasibility evaluation of potential concepts
A broad feasibility evaluation study of candidate canister solutions for the disposal of SF and vitrified HLW, involving the consideration of thick-walled 5 m long canisters for SF and 1.5 or 3 m long canisters for HLW, has been made by [5]. The assessment was based on the consideration of a wide range of factors under four generic headings: mechanical integrity, environmental damage (including impact on the geological barrier), fabrication and costs. The first three represent fundamental factors that relate to the requirements given in the previous section, whereas the cost factor was added because the selection of materials and concepts was deliberately very broad and it was judged that some insight was needed into project implementation costs especially where novel developments might be needed. The objective was to perform a broad-based evaluation of possible materials and design concepts, with a view to assessing the project risk of pursuing the various potential solutions over the next decades.
Canisters were assumed to be constructed from a number of candidate materials and the structural integrity of cylinders of the scale, suited to encapsulating SF assemblies and HLW, was evaluated using a failure-assessment diagram (FAD) methodology, see [5], without consideration of lid design or weld closure.
Carbon steel canisters with an appropriate wall thickness: (a) for the required structural strength, with a corrosion allowance, and (b) to ensure that the radiation level at the external surface is insufficient to influence corrosion resistance, were judged to provide a possible solution. There is well-established manufacturing and fabrication experience with parts of the required size, and the associated costs are well known. The main concern with this solution relates to the evolution of hydrogen as a byproduct of the corrosion process, although the assessments of hydrogen transport in Opalinus Clay suggest that the rate is low enough so that there would be no impact on the rock as a radionuclide transport barrier [6].
There is, nonetheless, some incentive to adopt a canister solution with negligible hydrogen production during corrosion, as the residual uncertainties in this area could then be avoided. Furthermore, such a design can lead to significantly longer canister lifetimes. The use of copper as a coating, with structural strength provided by steel or cast iron internals, was considered to be a viable solution. This concept might overcome uncertainties associated with creep of copper in the KBS-3 design, should these prove to be problematic [7]. The development of the copper coatings is continuing [8].
Disposal canister designs based on carbon steel clad with titanium or nickel alloy were also evaluated. Owing to the high corrosion resistance, the long-term rate of hydrogen production is low enough to keep hydrogen dissolved in repository pore-water and no gas phase would form. While coating with both materials is feasible with existing manufacturing and fabrication technology, there are uncertainties associated with long-term environmental damage resistance due to insufficient existing long-term data. Moreover, material and fabrication costs are relatively high. The mechanical integrity of solid titanium and nickel alloy canisters were also evaluated.
The potential for the use of ceramics in constructing disposal canisters was also considered, as some of these materials are extremely corrosion resistant and no hydrogen is produced when they corrode. A number of possibilities exist, including the use of ceramics such as Al2O3/SiO2 or SSiC (sintered silicon carbide). For such solutions, significant concerns associated with mechanical integrity, large-part manufacture and final-sealing feasibility could conceivably be overcome for the smaller diameter, shorter HLW canisters with appropriate development activity and financial investment, but it is highly unlikely that large SF canisters could be constructed with such materials. It is also debatable if the very high level of funding which would be required is justifiable for the relatively low number of HLW canister units ultimately needed, in particular when the successful outcome of such a research and development activity would be by no means assured, and if recommended further investigations confirm that there are already acceptable alternative solutions. In principle, other possibilities do exist, including coating or cladding a carbon steel sub-structure with a highly corrosion-resistant material.
Copper-coated canister concepts
In principle, a copper-coated canister could be based on either a cast iron structure or a forged or cast carbon steel structure as a substrate for the copper coating. In both cases, the necessary structural strength is provided by the internal structure and the required long-term corrosion durability is provided by the copper. The following discussion focuses on the carbon steel concept which is further advanced.
The basic design concept being pursued is a carbon steel cylinder electroplated with copper with lid and weld joint designs for the steel that would satisfy both the requirements to avoid post-weld heat treatment (PWHT) and to withstand the structural loads in a geological repository. The finished weld region would be copper coated using cold spray technology. A design study for a carbon steel canister [4] showed that a thick-walled canister (140 mm) 1 with an outside diameter of 1050 mm with a flat lid and a full section weld would remain structurally sound over its design lifetime under the expected loading conditions in a repository in Opalinus Clay. Nonetheless, a concern with the design is that a thick section weld would require PWHT which would be time-consuming and result in slow throughput, which would be impractical for hot cell operations, and may also heat the waste to unacceptably high temperatures.
An alternative approach, of using a different lid geometry with a reduced weld penetration, has been evaluated in several studies. Because the copper coating would provide corrosion protection and thus a corrosion allowance for the steel is not required, the steel wall thickness could be reduced to 120 mm, which is sufficient for the structural load. Studies of copper coating technology and corrosion are being performed in a cooperative programme with NWMO (Canada) [8] and are not discussed here. Design studies involved structural analyses of canisters with various lid geometries and welded joint configurations as well as evaluation of a number of welding methods. Three potential lid and welded joint designs were initially evaluated based on structural and manufacturing considerations. These were an elliptical lid with an axially oriented welded joint, a torispherical lid with an axially oriented welded joint and a hemispherical lid with a radially oriented welded joint. Based on the structural analysis and welding evaluations, it was concluded that a hemispherical lid with a radially oriented weld joint would be preferred [9]. The work, described in detail below, focused on hemispherical lid design concepts with the aim to optimise the weld depth and assess different weld process options.
Weld closure design
The main objectives of the structural analysis investigations of canisters with a hemispherical lid [10] were to
calculate the stress intensity factors at the tip of the root of the weld for different welded joint designs with various weld penetration depths, calculate the limit load causing plastic yielding for the same geometries and use these data in engineering critical assessments (ECAs) of the closure welds in order to determine their acceptability and safety margin under design load.
Several alternative designs with a hemispherical head and weld penetration depths of 25, 32 and 50 mm were proposed in which the weld depth and root geometry were changed to allow the investigation of the effect of these details on structural integrity. Included in the analysis was investigation of the effects of residual stress and weld flaws.
Material properties
The steel grade assumed was A106 Gr. C. From ASME II, Part D, Table Y1 [11], the yield stress and ultimate tensile stress (UTS) at room temperature for this material are 276 and 483 MPa, respectively. The Young's modulus and Poisson's ratio are 207 GPa and 0.3, respectively.
The models analysed for the calculation of stress intensity factors caused by the primary design load or the residual stress from welding assumed that the behaviour was linear elastic. The models analysed for the prediction of the plastic yielding load assumed that the behaviour was elastic – perfectly plastic.
Engineering critical assessment
The ECA procedure is based on the [12] FAD. The coordinates L r and K r of the assessment points were calculated for each primary load factor by dividing the factored primary load by the load causing plastic yielding of the ligament between the flaw tip and the outer surface for L r, and for K r by dividing the highest value of the stress intensity factor K eq calculated along the circumference by the assumed fracture toughness K mat. The UTS value, combined with the yield stress value, was used for the calculation of the FAD curve. The FAD curve defining the domain where the flaw is acceptable for the design load assumed, is defined in [12] Option 1. In terms of fracture toughness, the value assumed for the FAD was 80 MPa√m. This represents a conservative lower bound value for a HAZ in the as-welded condition assuming hydrogen embrittlement. The base material specification requires a minimum of ∼100 MPa√m.
In the study, finite element analysis was used for the calculation of the data necessary for the ECA. Stress intensity factors, at the tip of the root gap or the tip of an assumed welding defect, were calculated directly by Abaqus [13]. The limit load was obtained from elastic plastic analyses and defined as the load necessary to cause yielding of a ligament extending from the tip of the flaw to a free surface. It is worth noting that the term ‘plastic collapse’ refers to yielding when the local stress field is tensile. In the cases analysed, the crack tip stress field was mainly compressive, due to the external pressure applied. Therefore, the limit load corresponds to a failure mode caused by localised plastic buckling. For this reason, the condition where the limit is reached is referred to as ‘plastic yielding’.
Geometry and mesh
The analysis and results below deal only with a weld depth of 25 mm and the weld geometry is shown in Figure 2. The outer diameter of the canister was assumed to be 1020 mm, and the wall thickness was 120 mm.
Sketch of preferred welded joint design, with radially oriented root gap; dimensions in millimetre [10].
The finite element model for the design in Figure 2 is shown in Figure 3. The design in Figure 2 results in a compressive stress field at the root tip. Failure occurs mainly due to localised buckling when the applied pressure is high enough to cause yielding of the ligament near the weld.
Loading
At the outer surface of the cylindrical part of the canister, an anisotropic design pressure distribution was assumed, where the external pressure was 22 MPa on the vertical axis, and 29 MPa on the horizontal axis. The angular variation of pressure followed the sinusoidal function below, where 
is the angle measured from the horizontal axis:
Structural analysis results with and without residual stress
The effects of residual stress on the stress intensity factor, at the tip of the root gap and on the plastic yielding load, were determined by applying to the crack faces a pressure equal to the yield stress of the parent material (276 MPa). The width of the surface, where the yield stress was applied, was 7.3 mm (Figure 4). This value was obtained following recommendations in BS 7910 Annex Q [14] and depends on the estimated heat input from welding and material's yield stress.
Surface defined for application of yield magnitude pressure [10].
An example of an FAD for a 25 mm weld depth penetration is shown in Figure 5 for both anisotropic and isotropic loading conditions for cases with and without residual stress. A consequence of the root being loaded in compression is the improved ability to resist the plastic yielding obtained when the residual stress from welding is taken into account. The residual stress assumed here was of yield magnitude, in accordance with recommendations from [12]. In the cases investigated, residual stress influences the predicted plastic yielding load. The crack tip load caused by the residual stress is tensile and delays the compressive yielding of the ligament, so that the predicted plastic yielding load is increased by 25-30% depending on the cases investigated (Figure 5). Because the loading is compressive, accounting for yield magnitude residual stress is not conservative. A better estimate of the load, causing plastic buckling of the ligament, could be obtained from finite element analysis of the welding process and residual stress measurements. If this estimate is not available, the assessment of the design in Figure 2 would be more conservative if the residual stress is omitted.
Structural analysis results for the case of a weld flaw
The behaviour of the closure weld was examined for a weld flaw present in the HAZ of the root pass in the area of maximum stress. The size and geometry of the flaw were based on the assumed reliability of detection of the proposed post-weld non-destructive inspection methods. It has been suggested [10] that automated ultrasonic testing of welds in ferritic steel in a thickness of between 3 and 80 mm can be considered to reliably detect planar flaws of 1.5-3 mm in height with a length of 10 mm. For surface inspection using eddy currents, a detection capability of 3 mm height by 15 mm length is indicated. For these reasons, it was considered that the maximum flaw size that could potentially exist without detection would be of the order of 3 mm in height by 10 mm in length with just-buried planar flaws being the most damaging. In addition, the orientation is most severe as the local shear stress in this region develops the maximum tensile principal stress. Also, being located in the as-welded HAZ, it would be most critical due to the assumed level of fracture toughness being lowest in this region.
The geometry with the 25 mm weld penetration depth was used to investigate the effect of different material and loading conditions. The model of the weld flaw is shown in Figure 6. Results for the weld flaw case with and without residual stress are shown in Figure 7.
Position and geometry of the assumed semi-elliptical defect in the HAZ near the root, for the case where the effect of this defect on the position of the assessment points was investigated [10]. FAD for a weld defect near the radially oriented root assuming 25 mm weld penetration and anisotropic swelling pressure varying from 22 to 29 MPa, with and without residual stress [10].
When comparing the case without residual stress, for the anisotropic swelling pressure, the safety margin decreases from 1.5 to 1.2 (Figure 7). Similarly, from the FAD in Figure 7, the safety factor calculated including residual stress would decrease from 1.8 to 1.4 when the defect is assumed. The model generated for the case with the welding defect did not include the residual stress, and the graph with residual stress is an estimate based on the results of the previously analysed models; the data were not obtained directly from finite element analysis of this specific case. However, if a better estimate of the safety margin is needed, the model could be analysed accounting for residual stress affecting the crack tip loading and ligament yielding.
Weld technique assessment
In order to establish the basis for proceeding with design development work for the design shown in Figure 2, a review of candidate welding processes was undertaken [15]. The objectives were to evaluate processes according to a number of criteria outlined below and select at least two processes worthy of further consideration and development for the current canister design. Electron beam (EB), laser beam, friction stir welding (FSW) and arc welding processes (narrow groove tungsten inert gas (TIG) and metal active gas (MAG)) were assessed.
A set of 25 criteria were employed for evaluating the weld processes. Of these, 16 are identified as being of greatest importance. These key criteria are
The technology readiness level (TRL) of the process. Whether the process can be carried out automatically and remotely. Whether processing equipment can be exposed to radioactive environments. Whether associated process monitoring equipment can be exposed to radioactive environments. Whether the welding equipment is prone to breakdowns or if consumables need frequent replacement and, if so, the downtime associated with carrying out repairs. Whether the process can be operated in a 2G welding position. Whether the processing equipment can be rotated around a fixed canister/lid assembly in that position. Whether the process is suitable for application to the joint design. Whether the process is susceptible to residual magnetism. Whether, for multiple pass processes, any inter-pass cleaning is needed. Whether relevant welding procedure qualification standards for that process already exist. The likelihood of that given process producing weld defects. The likelihood of that given process also producing weld defects when it starts and/or stops. Whether that process requires the joint to be preheated. Whether that process is likely to produce an unacceptably hard weld. Whether that process is likely to produce high magnitude tensile residual stresses at the surfaces of the weld zone. Whether the process requires a filler addition. Whether the process requires gas shielding. The service requirements that the process requires. The suitability of the finished weld bead that the process produces for inspection and coating. The requirements of the process on joint cleanliness. The requirements of the process on joint fit-up. The estimated welding time of the process, as long as that time is <12 hours. The likely weld strength produced by the process. The likely weld toughness produced by the process.
Minor criteria, which have been considered nonetheless, are:
Different scores were assigned to these criteria, such that a score of 0 means that a given candidate process does not meet a given criterion at all, and 10 means, conversely, that the criterion is entirely satisfied. Different weightings were also assigned to these criteria, based on the importance of each criterion to the specific application. Key criteria have been assigned weights of ≥7, while minor criteria have been assigned smaller weights. On the basis of the total weighted score output from this evaluation for each candidate process, the two joining processes with the highest totals have been proposed as candidates for the closure welding of the new canister design.
Summary of overall scores for candidate welding processes.
Scores, weighted scores and overall scores for EB and FSW.
The scores presented in Table 1 suggest that the EB welding process is most suited to the making of the closure welds in the current Nagra canister design. Among the principal reasons for this are the maturity and simplicity of the process. However, EB welding requires a vacuum or a reduced pressure atmosphere to operate, and while the vacuum engineering required to achieve the operating pressure necessary is very simple, this requirement could still be construed as a major disadvantage. Conversely, operation in vacuum does eliminate the gas shielding requirement and provides fume control.
Table 1 also suggests that the next most suited process is FSW. As shown in Table 2, FSW scores highly primarily due to its single-pass nature, low propensity to weld defects, production of a hot worked fine grained weld zone microstructure with potential benefits in terms of residual stresses and corrosion behaviour, immunity to residual magnetism, and lack of need for preheat. Nevertheless, the weld depth range required is beyond current practice. Further equipment and process development work is, therefore, needed to achieve the 25-32 mm penetration depth range required. Consequently, FSW for this application is, at present, at a low TRL. In addition, information on tool exit hole management, tool life and recovery strategy for tool breakages requires development. FSW for this application currently lacks a standard(s) for procedure qualification. It is probable that the FSW process will develop to a point at which this application can be reliably accommodated, but this is not the position at the current time.
Table 1 indicates that narrow gap TIG welding would rank third among the processes considered. Nevertheless, arc welding methods currently appear to require more operator intervention to achieve reliable performance, e.g. inter-pass cleaning or inter-pass setting, and have scored lower when it comes to remote operation, as a result. Finally, for laser welding, a greater amount of procedure development could be required than for EB welding, given the greater maturity of the latter. The review showed that laser welding in vacuum or at reduced pressure and/or multi-pass welding using a wire fill will offer substantial advances in the state of the art in the future, not least in terms of the penetration depths that can be achieved. Nevertheless, these benefits have only recently begun to be commercially proven.
For the EB and FSW processes, further evaluation of structural performance was done considering defects typically occurring as consequence of these weld processes. Acceptable performance was found in the case of such defects [15].
Conclusions
A feasibility evaluation of canister designs and materials indicated that a Cu-coated carbon steel canister is a promising solution for the disposal of SF/HLW in a deep geological repository in Opalinus Clay. Preliminary canister and welded joint design studies have led to the proposal of a hemispherical head design with a root gap in the radial direction and a weld penetration of ∼25-32 mm. This design is preferred due to the weld root being closed by the external pressure on the head. The dominant failure mode is localised buckling in the ligament between the root gap and the external surface. Assuming an anisotropic swelling pressure varying between 22 and 29 MPa, the safety margin is 1.45 for 25 mm weld penetration, when the residual stress is omitted. When a yield magnitude residual stress is included in the analyses, the safety margin increases to 1.85. Assuming a 3 mm × 10 mm root defect in the HAZ results in a reduction in the safety margin from 1.45 to 1.2 without residual stresses. When yield magnitude residual stress is assumed, the defect may decrease the safety margin from 1.85 to 1.4, based on a semi-quantitative estimate. Because of the external pressure applied to the cap, accounting for the residual stress increases safety margins, so assuming residual stress is absent is conservative. The reduced weld penetration is considered to be beneficial, as residual stresses from welding are lower and PWHT is not likely to be needed. A weld process evaluation indicated that EB welding appears to be preferred unless vacuum or reduced pressure deployment is prohibitive for any reason. However, advances in some of the evaluated processes could result in a need for re-evaluation of the suitability of these processes for this application in the medium-term future.
