Improving the Accuracy of the LHC Superconducting Main Dipole Circuit By-Pass Diode Model Through Closed-Loop Optimization

The Large Hadron Collider (LHC) includes eight main dipole circuits, each containing 154 superconducting magnets. We are carrying out simulations of the electromagnetic thermal behavior of these circuits, together with experimental tests and continuous monitoring, to ensure their safe operation and to further improve our understanding. Accurately modeling these eight circuits is very challenging. This is due to the interactions between the 154 magnets in each circuit and due to small variations in the physical properties of the magnets, e.g. due to component tolerances which introduce numerous unknown parameters into the model. In this contribution, we propose a method to optimize these large parameter sets efficiently, using a novel software module that combines our previously in-house developed circuit simulation framework with Dakota, an open-source parameter optimization toolkit. We demonstrate the effectiveness of this approach by optimizing a new simulation model of the dipole magnet by-pass diodes. Improving the accuracy of the model may in the future allow the identification of magnets and circuit components with non-conform behavior, and the reconstruction of failure cases for which no measurement data are available.