Assessing Vircators’ Reliability Through Uncertainty and Sensitivity Analyses Using a Surrogate Model

A Framework for Peak Power Exceedances of High Power Microwave Radiators Applied to a Vircator Surrogate Model

“…Recently, a surrogate model of the Vircator [18] has been proposed in order to perform stochastic simulations at reduced computational cost [17]. We describe in what follows the simulation framework and the related uncertain parameters considered in this study.…”

“…The domain of validity of the surrogate model, described by the combination of Eqs. (1) and (2), is studied and defined in [17,23], showing applicability in a region of the variable space where particularly the optimal response of the Vircator is achieved. The particularity of this surrogate model is that it was adjusted to obtain lower error as the Vircator power limit was reached.…”

“…(1) and (2) is time-varying, and the analytical solution of the differential equation for v(t), which considers both the Vircator and the supplying network, cannot be solved in closed form. More details are presented in [17]. Because in a Vircator, the diode current is given by Child-Langmuir's law i(t) = Kv(t) 3/2 [27,28], the diode (load) impedance can be defined as a function of the voltage or diode current as…”

“…The modelling and simulation of Vircators can be performed by means of different simulations tools such as XOOPIC [15] and CST-PS [16] which are 2.5D and 3D particle-in-cell simulations, respectively. Interestingly, recently, a stochastic analysis of a Vircator, based on a surrogate model, has been proposed allowing the validation of the reliability of the system by studying RVs considering mechanical tolerances and pulsed power source variability [17]. Among many other advantages, the surrogate model allows for an efficient simulation at a reduced computational cost compared to XOOPIC and CST (surrogate: 19 × 10 −6 s, CST-PS: 6.8×10 4 s, and XOOPIC: 7.9×10 2 s -results obtained for the same Vircator configuration and on the same computational platform) [17] enabling different statistical analysis.…”

“…Interestingly, recently, a stochastic analysis of a Vircator, based on a surrogate model, has been proposed allowing the validation of the reliability of the system by studying RVs considering mechanical tolerances and pulsed power source variability [17]. Among many other advantages, the surrogate model allows for an efficient simulation at a reduced computational cost compared to XOOPIC and CST (surrogate: 19 × 10 −6 s, CST-PS: 6.8×10 4 s, and XOOPIC: 7.9×10 2 s -results obtained for the same Vircator configuration and on the same computational platform) [17] enabling different statistical analysis. Applying sensitivity analysis techniques, such as Sobol and Morris indices, which only focus on general tendencies, it has been demonstrated that the anode-cathode gap has the highest influence on the power response, in the case of mechanical tolerances, while the charging voltage of the pulsed power source has the most significant influence as a volatile property of the electrical components.…”