Elsevier

Fusion Engineering and Design

Volume 146, Part B, September 2019, Pages 2527-2533
Fusion Engineering and Design

Investigation of divertor movement during disruptions in ASDEX Upgrade

https://doi.org/10.1016/j.fusengdes.2019.04.034Get rights and content

Highlights

  • Composition of the divertor assembly 1 in ASDEX Upgrade.

  • Measurement of displacements of the divertor assemblies over an experimental campaign.

  • Results of full 3D electromagnetic simulations regarding forces and torques on the assembly.

  • Measurement of friction coefficient between stainless steel and silicon nitride.

  • Results of transient structural analysis regarding displacements.

Abstract

The divertor serves as the main power exhaust of tokamaks. Hence the target tiles in the divertor must be carefully aligned to prevent leading edges which would result in higher power deposition and subsequent melting. The outer strike line in the lower divertor of ASDEX Upgrade is located on the assembly 1, which consists of the target tiles, the cooling plates and the support structure. Since the transition to the tungsten optimized divertor design of the divertor in 2014, it has been observed that the assembly 1 and the underlying frame are displaced over the course of an experimental campaign. The attachment of the assembly has been modified several times to prevent this displacement. However, a complete suppression of the movement was not achieved. The reason for the displacement is suspected to be due to induced currents and the resulting jxB forces during disruptions. This was investigated using a full 3D transient model of the ASDEX Upgrade coil system and a model of the assembly 1 with frame and vacuum vessel. The different assembly modifications and several current quench times were simulated, resulting in forces of up to 5.5 kN and torques up to 6.3 kN m. These forces were then used in a 3D transient structural model of the assembly to investigate the resulting displacements. It was found that displacements occur in all cases but they vary between 0.25 mm and 0.75 mm.

Introduction

The divertor is an important component of most fusion devices, as it serves as power exhaust for the machine. This is achieved by diverting the plasma outside the last closed flux surface into an outer and inner leg which are then guided onto special target tiles. The strike lines, on which the plasma hits the tiles, are characterized by heat fluxes of up to 15 MW/m² in ASDEX Upgrade [1]. For this reason the tiles are either made of graphite [2] or tungsten [3] as these materials can withstand these heat fluxes. The alignment of the target tiles is also of great importance as leading edges are exposed to higher heat fluxes than the tile surface. This can lead to severe damage or destruction of entire tiles. For this reason, the tungsten divertor in ASDEX Upgrade (AUG) is carefully aligned after every maintenance break to prevent leading edges and to ensure proper shading across neighboring divertor assemblies. The divertor of AUG was originally equipped with graphite tiles. Since graphite is not suitable as target material for reactors, and thus for ITER, due to hydrogen co-deposition, AUG was stepwise transformed to a tungsten first wall [4]. The current design of the lower divertor (Div-III [5]), with solid tungsten tiles on the outer divertor, was introduced in 2014 to expand the experimental capabilities of AUG.

Since this change it was observed that the divertor assembly 1, which holds the target tiles for the outer strike line, was displaced in radial direction during an experimental campaign. Fig. 1 shows the radial offsets of the assemblies with respect to their neighboring assemblies in each toroidal sector directly before and directly after the experimental campaign between February and July 2017, which stands exemplary for the behavior in the campaigns prior. The green line represents the status of the assemblies after maintenance and the red line indicates the positions directly after opening the machine. It can be seen that during the experiments displacements between 0 mm and 0.35 mm occurred. Depending on the direction of the displacement, this increased or decreased shading between the assemblies, resulting in leading edges, higher heat fluxes and partial melting of tiles.

As of 2017, the assembly 1 consists of a support structure (Fig. 2 (1)) which holds two cooling plates (2). A pipe (3) for the cooling water is attached to these cooling plates. The cooling water is fed through a flange (4a) into the pipe and exits the assembly through a second flange (4b). Both flanges are mounted to the support structure. Each assembly holds eight tungsten tiles (5), four on each cooling plate. A 2 mm sheet of Papyex (6) between the tiles and the cooling plate serves as thermal conductor. The tiles and the Papyex are firmly pulled against the cooling plates by clamps above (7a) and below (7b) the tiles. The assembly 1 is held in place by the two water flanges which are attached to an underlying frame and by two sockets (left (8b) and right (8a)) which are mounted directly to the vacuum vessel. The frame itself is also attached to the AUG vacuum vessel. Except the tiles and the Papyex, all components of the assembly 1 are made from stainless steel, as is the supporting frame.

The sockets of the assembly 1 have undergone a series of modifications due to installation space and to prevent movement. These different versions, in which no socket, one socket or both sockets are electrically insulated, are represented in the simulated cases.

The reason for this movement is suspected to be jxB forces resulting from induced currents during disruptions. Other reasons for the displacement like thermal stress or halo currents are not investigated since the temperature of the support structure is never elevated significantly and halo currents create a uniform force on the assembly, which would result an increased shading of one neighboring assembly and a decreased shading of the other one. The investigation of the eddy currents and their effect on the assembly are described in this paper. This includes the electromagnetic model to calculate the forces and torques on the assembly 1 and a finite element model to simulate the displacement of the assembly.

Section snippets

Electromagnetic model

This analysis was carried out in ANSYS Maxwell. A full 3D model of the AUG coil system was used for this investigation, consisting of 16 toroidal field coils, 5 ohmic heating coils, 12 vertical field coils and 4 outer and 2 inner control coils [6]. The space between the single coils and between the coils and the divertor assembly was set to be vacuum. All boundaries were regarded as insulated. The material of the coils was set to copper. The plasma was represented by a perfectly conducting coil

Finite element model

This investigation was done in ANSYS transient structural environment. The simplified 3D model of the assembly 1 one was used with the materials according to the cases. The water flanges were set to be fixed while the sockets had a friction boundary condition with the friction coefficients of stainless steel (0.2) or SiN (0.12) depending on the case. The normal force on the sockets was set to 60 kN. The time traces of the forces and torques were imported from the electromagnetic results. The

Conclusion and outlook

The movement of assembly 1 of the AUG divertor has been investigated using a 3D transient model for calculating electromagnetic forces and torques. The results pointed out current loops inside the assembly and through the surrounding structures which give rise to forces up to 5.5 kN. The friction coefficients between SiN and stainless steel were determined to be 0.18/0.16 for FN < = 5 kN and 0.12/0.11 for FN > 5 kN. With the forces and torques from the electromagnetic calculation and the

References (7)

There are more references available in the full text version of this article.
View full text