Completion of the first ITER toroidal field coil in Japan

The first ITER toroidal field coil (TFC) has been successfully manufactured by the Japanese Domestic Agency in January 2020. The ITER TFCs are the largest Niobium Tin (Nb3Sn) superconducting magnets in the world; each is enclosed in an austenitic stainless-steel case with a height of 16.5 m and total weight is 310 tons (Knaster et al 2008 IEEE Trans. Appl. Supercond. 18 495–498). A set of 18 TFCs will be installed around vacuum vessel to function as a plasma confinement magnet system. The responsibility to procure 18 TFCs and 1 spare coil is shared between European Domestic Agency and Japanese Domestic Agency (Bellesia et al 2020 IEEE Trans. Appl. Supercond. 30 4202205; Sborchia et al 2008 IEEE Trans. Appl. Supercond. 18 463–466). To hold a common magnetic and geometrical properties among all the TFCs, tight tolerances of sub-millimeter order are defined on each TFC. The fabrication of those massive magnets with such tight tolerances involved some major technical challenges. These technical challenges were solved by pre-assessment and process qualification through some qualification trials. As a result, techniques established to solve those challenges were implemented to the TFC manufacturing, leading to the successful completion of the first TFC. The details are described in the paper.


Introduction
In ITER, 18 toroidal field coils (TFC) are planned to be assembled around a torus-shape vacuum vessel as a plasma confinement (figure 1). They are D-shaped coils consisting of a superconducting winding pack (WP) enclosed in a massive austenitic stainless-steel case with a height of 16 m weighing * Author to whom any correspondence should be addressed.
Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. 310 tons. Niobium tin (Nb 3 Sn) was selected for the conductor material of TFC winding so that the WP can remain superconducting under its target peak magnetic field of 11.8 T and stores total magnetic energy of 41 GJ [1,2]. The operating temperature of TFC is around 4 K. TFC will experience large electromagnetic forces during operation, so the materials for TFC structures (TFCS) were selected to maintain the structural integrity in the ITER operation conditions [3]. However, for those, so called 'ITER grade' stainless steels, it is not easy to control welding deformation, which strongly affects the geometrical property of a TFC. On the other hand, the magnetic quality of a TFC is indicated by current center line (CCL). The CCL can be defined as the barycenter of all the 134 turns  constituting the WP of a TFC. To hold a common geometric and magnetic properties among all the TFCs, tight tolerances of sub-millimeter order are defined on each TFC. From those specifications, one can say that ITER TFC is the largest Nb 3 Sn superconducting magnets ever built and the completion of the 1st-of-a-kind ITER TFC is an important milestone in the fusion energy research field.
In 2018, we announced the completion of the first TFCS for ITER TFC [4] at the 27th IAEA Fusion Energy Conference. TFCS fabrication began with material development [5] and successfully completed with welding deformation control and welding quality control [4,6]. About a year later, the first Japanese WP was completed. The successful completion of the WP involved tremendous efforts not to damage the heat-treated superconductors during insertion into stainless steel structures or during insulation process [7,8].
Since then, the manufacturing of TFC proceeded to next phase, integration of a WP in a TFCS [9]. The integration is performed in a step-wise process; (1) insertion of a WP into inboard of a TFCS, (2) insertion of outboard of TFCS over the WP, (3) welding of TFCS and installation of inner covers to completely enclose the WP inside, (4) gap-filling impregnation of a space between the WP and the TFCS to finalize the positional relationship between them, (5) final machining of interface surfaces and holes to compensate for any misalignment of the WP and TFCS, and (6) final inspection (figure 2). In the final inspection, dimensional inspection (DI) and electrical tests are performed to verify the properties of a manufactured coil.
In integration of a WP to a TFCS, there are three main technical challenges (table 1): (i) satisfying the sub-millimeter tolerances on interfaces in spite of welding deformation during TFCS welding, (ii) impregnation of 4 mm narrow gaps  between a WP and a TFCS with high-viscosity resin [10], and (iii) precise positioning of a WP inside TFCS to optimize CCL positions of a coil while keeping sufficient gaps for impregnation [11,12] and tracing those CCL positions throughout the integration process and transferring the information to the TFCS [12].
In January 2020, the first ITER TFC in Japan was completed overcoming those challenges. In this paper, we report the successful completion of the first ITER TFC introducing the techniques developed to overcome those challenges. A harmonization technique developed to satisfy intercoil interface tolerances is described in section 3 while gap-filling techniques are presented in section 4. Then the WP position control technique and CCL tracing technique are presented in section 5.

Requirements on ITER toroidal field coils
As mentioned already, the function of TFCs is to generate a precise circular magnetic field in a vacuum vessel to contain the plasma within the region of interest. In order to realize the same field map in all the TFC's contribution to the ITER TF system, all TFCs must have a common magnetic property and precisely positioned. The requirements and technical challenges on TFCs are summarized in figure 3 and table 1.
The details are as following: • Challenge (i): to achieve precise positioning of TFCs, there are sub-millimeter tight tolerances on intercoil interfaces, where each TFC is connected to adjacent TFCs. However, welding deformations of multi-pass welding of stainless steel was difficult to control. Therefore, to satisfy the intercoil interface tolerances, welding deformations of TFCS were needed to be pre-estimated to consider the countermeasure as described in section 3; • Challenge (ii): there is a 4 mm minimum gap requirement between a WP and a TFCS to ensure complete filling of gap with resin. Since high-viscosity resin was chosen for gap-filling, qualification of the gap-filling process for narrow gap was necessary. Also, even if there were sufficient gaps during WP insertion, there was a risk that those gaps are not maintained when welding deformation of TFCS was large. Therefore, investigation of local allowable minimum gap was also performed as reported in section 4; and finally, • Challenge (iii): the magnetic property of a coil is characterized by its CCL. For ITER TFC, the CCL positions are evaluated at 8 locations spread over the D-shape; 3 locations over the inboard, 3 locations over the outboard, and 2 locations between the inboard and the outboard. The CCL is geometrically calculated as the barycenter of the 134 turns of conductor within the WP at each CCL evaluation location and the CCL information of the WP is transferred to TFC during the integration of the WP into the TFCS. The CCL positions are required to be within a cylinder of 2.6 mm diameter from the nominal location in inboard while 3 mm transverse deviation is allowed for outboard.  Tighter tolerance is defined for inboard since more accuracy is required closer to plasma. There were 2 key points to achieve CCL tolerances; (1) to precisely control CCL positions by aligning the CCL of a WP to the TFCS datum and (2) to accurately trace the CCL information until it was transferred to the TFCS references before gap-filling. To achieve tight tolerances given to TFCs, control of misalignments or deviations during assembly was necessary. The details are described in section 5.

Intercoil interfaces
Since the welding deformation has large impact on the positions and shapes of interfaces, extra materials were left on critical interfaces so that they can be machined to nominal shape after welding. On the other hand, leaving too much extra materials will lead to longer machining duration. To avoid unnecessary schedule delay, amounts of extra materials were minimized based on the results of welding deformation qualification trials and structural analysis.

Pre-assessment of welding deformation
Initially, structural analysis with FEM was performed based on the inherent strain parameters determined from the welding result of flat test plates. To improve the assessment accuracy, several welding qualification trials with partial full-size mockups of different portions of TFCS were performed. After each qualification trial, inherent strain parameters for applicable weld line were updated in the global structural analysis. From the analysis, we found that outboard shift (figure 4, A1) and inclination (figure 4, A2) occur during welding between inboard and outboard, while side-plate distortion (figure 4, B1) and D-shape distortion (figure 4, C1, C2 and C3) occur during inner cover welding. According to the pre-assessment results, the extra material amounts for each interface were decided. Then, during welding of actual coil, the TFCS shape was monitored with a laser tracker so that the welding pass order can be adjusted to control the deformation. The deformation results are summarized in figure 4. As a result of those efforts, the distortion on interfaces due to welding deformations were kept within the extra materials  except at the interface on outboard, called IOIS. Among the intercoil interfaces, only the IOIS interfaces were impacted largely by all of the welding deformations: outboard shift, outboard inclination, side plate distortion and D-shape distortion, while impacts of outboard inclination and D-shape distortion are negligible on others. As a result, the maximum distortion on IOIS exceeded the extra material amount by several mm.

Harmonization
The intermediate DI was performed after TFCS welding so that the final machining plan could be prepared while the TFC underwent the gap-filling process. In the intermediate DI, we confirmed that the IOIS interfaces were out-of-tolerances for a few mm on outer sides radially for the first coil. Since the IOIS interfaces are the connection structures between adjacent TFCs, the deviations on those interfaces impact the assemblability of TFCs around vacuum vessel.
Since the manufacturing responsibility of TFCs are shared between European Domestic Agency (EUDA) and JADA [2,3], we had a chance to compare the welding deformation results of two parties. In comparison, we found the common deformation trends. Focusing on these same trends, we concluded that the welding deformations observed were structurally unavoidable for TFCS welding. On the other hand, as long as the final shape of all TFCs are the same, the 18 TFCs can be assembled. Therefore, a remedial action was performed globally as harmonization action for all the TFCs under the guidance of ITER Organization. IOIS distortions were compensated by shifting nominal positions by 5 mm radially inward. As a result of the intermedial DI and harmonization, all the intercoil interfaces met the tolerances without any schedule delay.
In figure 5, the interface measurement results for subsequent TFCs after welding are shown. The positive deviations mean that there are some extra materials, which can be machined to nominal shapes during final machining. For the 1st TFC, the most of the measurement points, there were sufficient extra materials. There were some points below nominal; however, they were all within tolerances. With the extra  materials and the harmonization, the interface tolerances of the 1st TFC were met and assemblability was ensured.

Gap between a WP and a TFCS
The gap-filling is performed for the structural integration of a WP and a TFCS, and the resin composition was chosen accordingly to match the compressive strength and integrated thermal contraction [13,14]. However, the chosen resin composition resulted in high viscosity of 10 Pas for fresh mix and 26 Pas after 24 h from mixing even if the resin temperature was kept at 40 degrees Celsius, the mixing temperature of the resin. As prevention of void formation, gap-filling was planned with vacuum pressure impregnation and the minimum gap requirement to ensure the complete filling of the gap. Also, it was planned to inject the resin from the lowest injection holes. The valves for higher resin injection holes were opened as the resin level reached the hole positions to keep the fresh low viscosity resin at the resin front [9].
Once the resin injection was completed, the resin would be cured at 75 degrees Celsius for 24 h. After the cure, the TFC would be cooled down to the room temperature gradually. From the previous resin curing trials, it was known that the cured resin has a risk to crack during cooling due to contraction. If the cracked piece of the gap-filling resin crumbled, it might fall into the gap between the WP and the TFCS and damage the ground insulation of the WP. To prevent the resin crumbling, application of a fiberglass layer on the WPs was added to the plan [10].

Resin injection trial
As previously mentioned, there was a concern in a complete gap-filling with additional fiberglass layer, as well as a risk that a gap may not be maintained if a larger welding deformation occurs. Therefore, resin injection trials with narrow gap models were performed to establish a qualified gap-filling method ( figure 6). In the first trial with 2 mm gap model, resin injection was not achieved [9,10] with half-overlapped 0.25 mm thick fiberglass tape on WP model. In the trial, it was observed that the resin flow slowed down at the boundary to next fiberglass layer when fiberglass tape was wrapped against the resin flow direction (figure 6(b) (i)). It might be acting as filter for the filler in the resin. It was also noticed that as the resin flow slow down, the liquid portion of resin soaked into the fiberglass tape faster than the resin level increase, making the resin viscosity even higher. The additional fiberglass layer made the resin injection difficult by making the viscosity of the resin mixture higher and preventing the resin flow.
In the later qualification trials, different patterns of test conditions were selected to find the proper combination of the wrapping direction and the thickness of the fiberglass tape to improve the resin flow. The thinner fiberglass tape was selected to simply reduce the amount of soaking. Also, the fiberglass tape was wrapped in the opposite direction from the first trial. Then, the better smooth flow of resin was observed. Complete resin injection into 2 mm gap was achieved when the 0.13 mm fiberglass tape was wrapped in the direction of resin flow, half-overlapped as shown in figure 6(b) (ii).
Since the resin viscosity increase in time, there was a risk that the resin injection might not be resumed once resin injection stopped for some emergency situation. To qualify the resin injection procedure for such a case, each trial was performed over 2 days with 24 h pause to mockup the emergency situation. On the second day, resin injection was resumed with colored resin to visualize the resin flow. It took a longer time but the colored resin was able to push the old resin and completely fill the 2 mm gap model. Through the trials, the qualified gap-filling procedure for 2 mm gaps was established. The gap requirement of 4 mm minimum was relaxed to accept the local gap of 2 mm when 0.13 mm fiberglass tape was half-overlapped in the direction of resin flow.

CCL positions
As a technique to overcome the precise WP positioning, the target WP position within the TFCS was pre-assessed using the individual DI data of the WP and the TFCS [11]. Through the pre-assessment, the optimized CCL positions were determined so that both the CCL position requirement and gap requirement are met during the WP positioning inside the TFCS ( figure 7). Then, a technique to trace CCL positions throughout the manufacturing process is developed [12] so that the final CCL positions can be precisely evaluated. The traced CCL position information is transferred to the TFCS by measuring 8 of the reference points on the TFCS locating near the CCL evaluation locations (figure 8, CCL cross sections A-H) in the same coordinate system as CCL positions just before gapfilling impregnation. After the impregnation, the WP and the TFCS is a solid integrated body, and the CCL positions can be evaluated from the measurement positions of the 8 reference points using mechanical analysis with FEM model.

Pre-assessment
To achieve tight tolerances on CCL positions, pre-assessment of target WP position were performed. This target position was determined using individual DI results of a WP and a TFCS for each TFC, along with welding deformation expectation values; such as outboard inclination of ±3 mm maximum and side plate distortion of 4.5 mm maximum. The outboard inclination occurs during welding of inboard and outboard (figure 4, A2). From the welding qualification trial results and the tolerances defined on the outboard portion, the target control value for outboard inclination was set to ±3 mm maximum. The side plate distortion occurs during welding of inner covers (figure 4, B1). The welding qualification showed the maximum distortion of 4.5 mm. In the assessment, the gap between the WP and the TFCS was evaluated for several cases of outboard inclination (figure 7). From the gap assessment for the 1st TFC, it was found that the 4 mm gap is maintained for any case of outboard inclination within ±3 mm if the WP tilt is adjusted accordingly to optimize the CCL positions. Therefore, as long as the outboard inclination is controlled within ±3 mm, not only the CCL positional tolerances are met but also the gap requirement and the outboard tolerances are met.
During WP insertion, the WP position was monitored by laser tracker, and a fine adjustment of 0.1 mm was performed. As a result of the pre-assessment and the laser tracker monitoring, precise positioning of 0.3 mm to target position was achieved [12]. This target position of a WP within TFCS is an optimized position where the CCL deviations are minimum while keeping 4 mm minimum gap. After outboard installation, the gap assessment was repeated with the outboard inclination result to determine the amount of WP tilt adjustment. Then, after the WP tilt adjustment, the gaps between the WP and the TFCS were measured and compared to assessment result as verification of the adjustment before the installation of TFCS inner covers. The measurement results showed the validity of the pre-assessment. After TFCS welding, WP shape was corrected to optimize the CCL position deviation and the gap according to structural analysis [9,12].

Traceability measurements
After insertion of a WP, there are several steps until the WP is fixed inside the TFCS during gap-filling. Throughout those steps, WP positions and the shape were measured with laser tracker to update the CCL information so that any misalignments or deviations can be compensated during final machining to minimize errors. Also, each measurement is repeated several times to minimize the measurement uncertainties. Originally, a WP shape was considered to be independent of the welding deformations of a TFCS. Therefore, CCL information of a WP was going to be measured up to cover installation step. However, the concern of welding deformation impact on the WP shape was brought to attention by IO. Measurement up to gap-filling was added using resin injection holes. Since the locations of resin injection holes do not coincide with the CCL cross sections (figure 8), some additional technique was necessary to determine the final CCL positions. The final CCL positions were calculated through analysis of the final shape of the WP, using the measurement results at the resin injection holes and the determined structural properties of the WP. By those efforts, CCL positions for the 1st TFC met the requirements.

Completion of the 1st coil
The 1st ITER TFC was completed in January 2020. The final results of CCL positions, interface tolerances and gap between WP and TFCS are shown in figure 9. The graph on the top left shows the interface measurement results of the 1st TFC after final machining. All the intercoil interfaces met the tolerances. The graph on the bottom left shows the gap measurement results. 4 mm gaps were secured for the most of the regions and there was local minimum gap of 3 mm. The graph on the right shows the deviations in CCL positions with evaluation uncertainties. The values were all within tolerances and the deviations at the inboard were controlled under 0.35 mm. By satisfying those three requirements, the magnetic property and the assemblability of the 1st TFC were secured.

Conclusion
The world's first ITER TFC was manufactured in Japan in January 2020 (figure 10), satisfying the required tolerances to achieve its function. The developed techniques to overcome manufacturing challenges have been applied to subsequent TFCs. Now, ITER received 4 TFCs from EUDA and 3 TFCs from JADA, as our contribution to the progress in the ITER project.

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