Message from Technetics Group Regarding the Coronavirus
Research and Development on Metallic Seals for Neutral Beam Injectors Vessels

Research and Development on metallic seals for Neutral Beam injectors vessels


The ITER Neutral Beam (NB) injectors are used for heating and diagnostics operations. There are 4 injectors in total, 3 Heating Neutral Beam injectors (HNBs) and one Diagnostic Neutral Beam injector (DNB). The HNB injector beam source and the beam line components (BLC) creating the neutral beam are contained inside two welded metallic vessels, the Beam Line Vessel (BLV) and the Beam Source Vessel (BSV). All DNB components, ion source and BLCs, are contained in only one vessel.

The BLV lid provides the access for repair of replacement of the Beam Line Components and DNB Beam Source. The BSV lid allows the HNB Beam Source replacement. The lids are intended to be opened to allow replacement of the BLCs and Beam Sources.

In the current design the BLV and BSV lids have Primary Vacuum and Confinement Boundary joints that can be opened but are to be all-welded using a lip-seal arrangement. Welding/cutting works to open and close lids have to be completed under Remote Handling methods which are not technologically demonstrated as feasible yet. Lip-seal welds are not directly compliant by the design code used for NBI vessels (RCC-MR).

It has been then decided to investigate usage of use double metallic seals to ensure the confinement. This solution presents several advantages compared to lip welds: Double confinement of the vessels, do not limit the number of openings and allows the seal to be leak tested before the rest of the vacuum system is closed. A Research & Development activity was launched due to large dimensions of shaped metallic seals needed (9.4 meters x 3.3 meters) which are out of existing scale. This task has demonstrated manufacturability and leaktightness, both in normal and accidental conditions.

This paper describes the results obtained during the R&D tasks and applicability of this proposal to ITER needs. The concept of the seals design according to remote handling operations is also presented in this paper.

I. Introduction

A specific work has been carried out to develop a solution ensuring confinement of the ITER (Nuclear facility INB-174) Neutral Beam Injectors (NBI) vessels using metallic seals. This development was performed in three main activities conducted almost in parallel. The first was a Research and Development (R&D), task in collaboration with Technetics Company, to design, test and ensure manufacturability of the metallic seals. The second was the design of NBI vessels compliant with seals shape and requirements. The third activity was the conceptual design of remote handling systems demonstrating that replacement of the seals is feasible. This paper will present an overview of all these activities, describing the main points, and highlights on remaining issues to be solved.

II. Research and Development Activities

A. Need and objectives of a R&D activity

A R&D activity was required due to large dimensions of shaped metallic seals needed which are out of existing scale. Dimensions for Heating Neutral Beam (HNB) injectors’ Beam Line Vessel (BLV) lids are 9 meters x 3 meters and 4 meters x 4 meters for the Beam Source Vessel (BSV) lids. Dimensions of the Diagnostic Neutral Beam (DNB) injector lid are similar to those of the HNB BLV. These components are Protection Important Components and Safety Important Component class 1.

Besides the manufacturing capability demonstration of such large shaped seals, the R&D program aimed to demonstrate that sealing is ensured during normal and accidental situations and compatible with remote handling maintenance operations:

  1. Ensure sealing performances required for ITER first confinement and vacuum barrier during normal operations: Helium leak rate for HNB or DNB is 5.10-8 atm.cm3.s-1 (5.10-9 Pa.m3.s-1) for the 9x3 meters HNB/DNB top lids closure. This is considered as the most critical shape because of the 9meters length
  2. Ensure confinement performance in accidental condition: Maximal leak rate for one HNB lid is 0.6 Pa.m3.s-1 and for DNB lid is 0.3 Pa.m3.s-1
  3. Sealing surface quality must be maintained if seals need to be replaced to avoid machining closed to vacuum facing surfaces. For this part of the study, we checked if a permanent deformation of the seal track occurs during the tests or may occur according to Finite Element Analysis (FEA). No repetitive mounting is planned
  4. Seals must tolerate top lid deformations. To give data to answer this question, spring back is measured for each seals type. Radial motion effects are not considered.
  5. Assess the required tightening force to close the lid. Optimal linear load versus seal deformation is determined for each type of seal.


HELICOFLEX® spring energized shaped seals design are manufactured, considering different sealing lining material and different stiffness, to find the right balance between sealing performances, required compression loads and useful spring back.

B. Seals design

In a first approach it was decided to test the behavior of different seals design on a reduced shape seals (618mm x 300mm). A FEA is used to demonstrate representativeness of these mocks-up regarding the final scale one seals.

Test aimed to:

  • Make decision on the best suited shape seals design
  • Check their ability to reach the required sealing performances
  • Check the seal track after each test to evaluate the flanges marking risks.

Two lining materials have been selected for seals: Silver and silver with special treatment improving spring back. For each type of seals two stiffness of inner spring shape have been selected in order to have different tightening force. Several seals have been manufactured for each configuration (see table 1). Seals diameter is 10.2mm.

C. Tests results

Each type of seal has been tested within the following conditions for normal operations:

  • Room temperature
  • Testing tool material: 304L stainless steel
  • Inner pressure for normal operation test: <10-4Pa
  • Measurement length: 15minutes

Leakage criterion:

  • Normal operation: 3.3 x 10-9 Pa.m3.s-1

The leakage criterion has been scaled from those corresponding to seal of 24m long (9mx3m) given in first paragraph II-A to the 1.6m length of the tested seals.

Results of tests are summarized in table 1 here below

Surface marking measurements have been performed after each test in order to evaluate the “damage” on the initial roughness of the sealing track. Indeed the surface can be damaged because of the contact pressure on it due to the high compression force required to deform the seal. The leak rate is strongly correlated with the surface roughness: The higher the roughness, the lower the leakage rate is. The recommended Ra must be as close as possible of 0.8 µm.

Roughness measurements after tests reveal that no surface collapsing is observed under the sealing track where seals/flanges contact pressure is the greatest. However, the flanges surface have been regularly cleaned and buffed to remove silver deposit.

The leak tightness of seals has also been assessed in accidental conditions. This was necessary to ensure that confinement requirements are met and gave information for maximal spring back of seals in order to design the NBI vessels accordingly

The seals have been tested based on following accidental conditions:

  • Inner pressure for accidental case test: 2 bar absolute
  • Leakage criterion: 2 x 10-2 Pa.m3.s-1 corresponding to a pressure drop of 8 mbars/minute for 1.6m long seal. Confinement requirement will be measured based on pressure drop.

Considering the relation between pressure drop and leak rate, the pressure measurement performed suggested a corresponding leak rate smaller than 3 x 10-3 Pa.m3.s-1 under accidental situation. Thus, leak rate meets confinement requirements for accidental loading.

Removing the load progressively reveals a significant pressure drop when aperture reaches 0.25 mm, thereby determining the maximum allowed deformation of flanges in accidental situation.

D. Scaling analyses

FEA analyses have been performed in order to assess behavior of HELICOFLEX® seals. Two models have been checked. The first was built to represent a straight part when the second model represented a curved part with a radius of 110mm. Indeed two models are necessary because the curvature increase the mean distance between each internal coil’s turn, which decreases the spring stiffness.

These two models will be used to compare the calculated linear load between these two parts of the seals and compare it with measured force during test. See in figures here bellow example of results for medium silver seals.

In order to estimate the risk of marking the surface and damage the flange at full scale a numerical analysis was performed assessing the contact pressure distribution on flanges after compression. See results in the figures below.

The analyses revealed that contact pressure on flange can reach 450MPa at the top of each inner coil.

To assess the risk to affect the flanges material by plastic strain, an analogy with the Hertz’s theory of cylindrical contacts was considered. The maximum shear in NBI flanges reach 0.3 x 450 = 135 MPa. The maximum shear stress is significantly smaller than 180 MPa as specified in RCC-MRx. No significant plastic strain of flanges will occurs with such contact pressure.

The numerical analyses performed are relevant to predict linear load values and exhibit a slight under-estimation of the seals total spring-back. It does not reveal any difference between curved and straight parts which demonstrate the mock-up size relevance regarding the scale on product. The simulation also provides some evidence that no plastic collapse of the flanges occur under the seals contact track.

E. Manufacturability

It was also necessary to assess the manufacturability of such a large shaped seal. Indeed the proposed design for ITER consists in a double seal arrangement mounted on a spacer. The spacer is necessary in order to handle the seal arrangement without touching the seals themselves with the risk to damage them.

The first manufacturing tests were performed on 3m x 2m seals assumed as representative of the full scale seal.

A main difficulty encountered was the bending of the inner seal to get the proper radius and keep it attached on the spacer.

The manufacturing of the spacer was also investigated in order to ensure correct mounting of seals, to reduce the overall weight, to ensure required stiffness and how to assemble all part of it during the manufacturing process.

After several trials first 3m x 2m prototype has been realized.

A full scale double seal has been manufactured since these first trails.


Some design works have been conducted in order to assess the impact of implementation of seals on the injectors’ flanges. These tasks have allowed cross checking feasibility regarding requirements and constraints to be respected. Requirements and constraints are the dimension of the seals and spacer, the number of bolts and required tightening force to compress the seals and, finally, the maximal relative deformation acceptable of vessels and lids flanges.

A. Design

The first task was to implement the maximal number of bolts all around the vessels flanges. They must be compliant with the required tightening force and the gap between each shall be the minimum possible. Indeed the closer the bolts are next to each other the better the compression of the seals is. Bolts must also be as close as possible of the seals in order to optimize the compression of the seal. The seals rigidity is very high and tends to deform the flange and therefore not compress the inner seal enough. The applied force must be strong enough on the two seals to deform them enough and ensure the confinement.

In a second step it was necessary the check the manufacturability for the seals implementation. The metallic seals required high manufacturing level and very tight tolerance. These constraints are mainly applicable on the groove as shown in picture below.

Several manufacturers have been contacted in order to cross checked that the required tolerances are achievable. And feasibility on a 24m long groove has been confirmed.

B. Analysis

Starting from the typical bolting arrangement some mechanical analyses have been conducted assessing the design in two main loading cases:

  • Normal operation: High vacuum (1x10-6 Pa) inside the vessels.
  • Accidental case: pressure increase up to two bars absolute inside the vessels.

For all analyses presented here the linear for one seal the tightening force of 270kN/bolts gives a total load of 55.1MN to compress the 24m long seals arrangement.

The analyses conducted on the typical arrangement have shown how the system behaves in the two loading conditions and have highlighted the two following phenomena:

  • Because of the distance with the inner seal and the bolts it is not possible to get the required compression force only by applying the tightening force in bolts. Vacuum inside the vessel is necessary to reach the required deformation of metallic seals.
  • In case of accident the deformation of the top lid tends to open by rotating around the bolting point and compression of the seals is not ensured (see picture here below).

If the first phenomenon shows that reaching vacuum shall not be an issue, the second is more critical. The first approach to solve it was to reinforce the lid. But the possibilities are limited because the dead weight of the top lid is already close from lifting capacity of injector opening tools and it overall dimension are also at their maximal values regarding surrounding environment. Then it was necessary to investigate other solutions.

The main driver in others design proposal was to limit the relative opening between top lid and vessel and/or block the top lid rotation. To do so the following cases have been studied.

In this first case the top lid is linked to the vessel by four internal mechanisms on each side of the vessel linking the top lid with the vessel. They are locked after bolting. This solution gives good results but it would require the design of a moving mechanism inside vacuum and new penetrations on the vessel. The major drawbacks of this solution are that if this mechanism failed when closed then it will not be possible to open the injector with complex access reparation tooling as it is inside. This drawback could be compensated by designing dismountable penetrations allowing access inside the vessel.

In this second case the top lid is extended all around the vessel flange. This extension blocks the top lid rotation when deformed by internal overpressure. The maximal gap between both components respecting the maximal seals spring back for the inner seal is 0.2mm on each side. This gives a total gap of 0.4mm between top lid and vessel’s flange. This is too tight and this solution is considered as not acceptable. The risk of having top lid blocked on the vessel or not be able to replace it after opening is too high.

The case three was thought in order to avoid the issue a gap too tight between the vessel and the top lid extension. In this case 40 pins with a diameter of 48mm are placed all around the vessel flange. They cross the top lid extension and go inside the vessel flange to block the deformation of the top lid. This solution doesn’t work very well because pins are deformed and therefore do not stop enough top lid deformation.

The curves here bellow gives results of the relative opening of the top lid compared to the vessel flange in accidental case following 4 design solutions presented previously. Therefore it can be compared with the useful spring back of the seals and we can assess if leak tightness is maintain. And for each design case relative opening at the location of the outer and inner seals is given.

It clearly shows that only the design with internal locking system allows reaching confinement requirements both in normal and accidental conditions.


In parallel of the R&D activities and the assessment of a new design for the NBI vessels it was also necessary to assess the feasibility of the remote handling operations with metallic seals. The RH operation can be seen as three major different tasks: Transportation the seals in the NB cell, place the seal on the vessels, enter a new seal in the NB cell.

Conceptual design performed for each tasks is presented in the following paragraphs.

A. Transportation of the seals inside the NB cell

The metallic seals are positioned on the HNB BLV and DNB lid and on the HNB BSV. Handling is ensure by the remote handling crane running in the NB cell.

The first task performed on RH activity was to check that the seals can be transported inside the NB cell. If the rear lid seals do not present major difficulties, it is more complicated for the HNB BLV seals, especially for the HNB#1. It has to pass between the front pillar of the NB cell and the bioshield, which is tighter space in the NB cell for a 24m long component. This transportation study had also helped to define inclination of the seals during transportation and therefore the requirement for the handling tool. This study is summarized in the picture here bellow.

The smallest gap with surrounding equipment is 330mm. The smallest gap with building is 350mm. These gaps are assumed as enough to compensate manufacturing, construction and assembly tolerances.

B. Place the seals on the vessels

he second major operation to assess was the placement of he seals on the vessels. This operation is very delicate because the seal must be placed perfectly flat inside the groove to avoid any scratching which could damage the surfaces and therefore the leak tightness. This requires a specific tooling arrangement assuring flatness of the seal and smooth and controlled landing of it inside the groove.

The solution thought for it is that after transportation in the NB cell the RH crane the lifting structure is placed on several coarse alignment jacks placed all around the vessel. These jacks will then lift down the seals just 10mm above the groove. The final approach is ensured by a dedicated bolt placed in the lifting interface fixed on the spacer. Then all tools and interfaces are removed and the vessel can be closed. Major operations can be seen in picture bellow. As required forces to perform these tasks are low all actuators/bolts can be manipulated by the existing RH manipulator.

The operation for the vertical seals installed on the HNB BSV will be the same.

C. Bring a new seals in the NB cell

The last major point to be checked to ensure feasibility of the metallic seals solution is how to enter new seals in the NB cell. Indeed the seals, especially the 9m long seals are much longer than the other components where a 8m long maintenance task is foreseen. All others components are brought into the NB cell using this cask because NB cell cannot be opened to others rooms. However is 1m shorter than the metallic seals. Therefore it was necessary to think to a dedicated solution. In order to limit cost impact of the solution, the idea was to find a solution with existing tools with minor and cost effective modifications if necessary.

To realize it, a simple sequence was thought. It is realized in the following steps:

  1. Dock the cask to confinement door between NB cell and hot cell building. This allows using the cask as temporary confinement barrier.
  2. Bring the seals on the back of the cask.
  3. Open the rear door of the cask.
  4. Move the seals inside the cask.
  5. Install a temporary cover around part of the seals longer than the cask.
  6. Open the confinement door.
  7. Moved the seals inside the NB cell.
  8. Closed the confinement door.

Basic views explaining the here above process are shown below.

In case of removal of seals, it is not necessary to reproduce this process. The seals can be cut inside the NB cell and place in the normal cask before being treated in hot cell building.

All the RH operations tend to demonstrate that remote replacement of the seals is feasible. However two others points will require design task. The first is about the accurate closing of the top lid when new seal is in place. Controlled lowering and engagement with metallic seals is a critical point. Accurate guiding system must de designed. The last point to be looked concerns the reparation of a damage surface onto the vessels or lids. If a small imprint appears near sealing tracks a “scotch-brite” cleaning is performed. If a significant defect (scratches, marking) appears or if roughness is too small a buffing is performed to reach the 0.8 recommended Ra with parallel scratches. Today there is no existing automatic process. It is usually done by hands. An automatic system is under development by seals manufacturer.


The sealing of the NBI vessels with metallic seals is a complex task. The complexity is due to the dimensions of the seals which are currently out of existing scale, and the difficulty is increased by the requirement to replace them remotely. The complete R&D and design tasks performed to demonstrate feasibility of this solution has been described here. This allows demonstrating that this solution is feasible using existing tooling technology and in a cost effective manner in contrary of the baseline solution. However some points must be cleared before giving final conclusion. Test on seals with large dimension shall be performed to guaranty that.


The views and opinions expressed herein do not necessarily reflect those of the ITER Organization.



This article was presented at the 2015 IEEE Conference. Download the original article as a pdf.