Speaker
Description
The COMPASS-Upgrade tokamak will be a high field compact device (R = 0.9 m, Bt = 5 T, Ip = 2 MA, tflattop = 2-10 s) capable of studying diverse magnetic configurations at high plasma densities [1]. The small machine size, high B and relatively high heating power means that scrape-off layer widths are predicted to be small (λq~0.6–1.2 mm), and attached power densities up to 100 MW m-2 could be achieved [2]. Due to the lack of active cooling, such loading would result in rapid heating of the W divertor tiles up to 2000 °C within a fraction of the total pulse length.
In order to evaluate the expected response to such repeated loading under relevant heat and particle loading conditions, the Magnum-PSI device [3] was used to compare three candidate tungsten grades (ALMT, AT&M, Plansee). In order to design a heating and cooling cycle which could achieve the desired conditions a finite element model was created using COMSOL Multiphysics. Preliminary testing was carried out to better evaluate the unknown input parameters to this model: through-plane conductivity of the grafoil thermal interface material (TIM) HT-1220 and the heat transfer coefficient of the Magnum-PSI jet-impingement cooling. This was carried out via heat loading measurements on a dedicated test mock-up using five type-K thermocouples to provide data while the plasma power was varied from 1.5 – 6.5 MW m-2 and TIM layer number, vacuum gap diameter or applied compression was changed.
Based on this input a cyclical loading scheme was applied to nine 10x12x3 mm tungsten samples. Each of the three grades was heated for 100 cycles up to either 2000 °C, 1800 °C or 1600 °C, with a heating time of 4.5 seconds and a cooling period of 39.5 seconds. From inspection by SEM, despite the short total duration, recrystallization was observed in all samples. Additionally, for the 2000 °C samples, the central region displayed cannibalistic grain growth and strong plastic deformation. Surrounding this region heavy roughening was also observed, but no cracking was detected. All grades behaved similarly. Overall the results show that strong microstructural and mechanical changes should be expected, but that the tiles should survive without failure.
[1] P. Vondracek et al., Fusion Engineering and Design, 169, 112490, (2021)
[2] M. Komm et al., Nucl. Fusion, 64, 076028, (2024)
[3] H.J.N. Van Eck et al., Fusion Engineering and Design, 142, 26–32, (2019)