Noise limits on ITER plasma vertical stabilization system imposed by tungsten divertor monoblock thermal fatigue

Abstract

Abstract Tokamak plasmas with vertically elongated cross-sections are unstable due to vertical displacements. Feedback stabilization of the ITER plasma current centre will be performed using the speed of its vertical displacements, dZ/dt, as input. Inevitable noise in the diagnostic signal for dZ/dt leads to “noisy” components in the feedback voltage and current in the stabilizing coils. This leads to noisy components in the vertical position of the plasma current centre and the divertor strike points. For burning plasma conditions on ITER, these strike point displacements may be a concern for thermal fatigue at the water cooling interface of the tungsten monoblocks constituting the divertor targets. A study is presented here in which this concern is examined for the first time. The baseline 15 MA scenario (fusion power of 500 MW with a 500 s flattop and fusion power gain, QDT = 10) is simulated with the DINA code, assuming low frequency noise in the dZ/dt diagnostic signal. The noise has uniform spectrum with a given root mean square (RMS) value ( =0.6 ms−1 or 0.2 ms−1) in the frequency band (0, 1 kHz). The results of these DINA simulations are combined with dissipative divertor plasma solutions obtained with the SOLPS-ITER plasma boundary code to provide a time dependent divertor target heat flux density profile. The latter is then imposed on a finite element model of the target monoblocks to assess the temporal evolution of the 3D temperature field in the block, including the Cu-W and Cu-CuCrZr joints at the cooling interface. These joints represent the points of the Cu and CuCrZr materials that see the largest temperature changes and are thus at greatest risk of failure. To evaluate an acceptance criterion on the non-cyclic and non-uniform thermal loads at the joints, an approach is developed which combines a Rainflow counting technique with Palmgren-Miner’s rule for fatigue accumulation. Analysis of the temperature evolution at the Cu-W joint shows that the RMS value of noise ∼0.6 ms−1 is unacceptable from the point of view of thermal fatigue for the expected total exposure time under burning plasma conditions that the first ITER divertor must survive. The lower value ( ∼0.2 ms−1) is found to be acceptable. A subsequent parametric study concludes that ≲ 0.44 ms−1 is just consistent with fatigue lifetime for the prescribed divertor power flux density profile, but should be kept lower than this to allow for some margin. Regarding the monoblock surface temperature, the natural power spreading caused by the separatrix movements is found to be beneficial from the point of view of recrystallization. For the level of noise in the dZ/dt diagnostic required to stay below joint fatigue limits, the average surface temperature on the most loaded monoblocks is reduced by ∼100 °C compared to the case with stationary strike points and the amount of time spent at this temperature at any one block by over 80 %.