Tether Discretisation in Airborne Wind Energy Systems: Convergence and Computational Cost

In time-marching simulation of airborne wind energy systems (AWESs), the tether is usually discretised into a series of lumped masses (i.e., nodes) connected by spring and damper elements. There is no consensus on the appropriate number of nodes for AWES applications, although it is known that using more nodes than necessary is computationally costly. In this work, we examine the trim solutions of a fixed-wing aircraft connected to the ground by a 500 m tether. A successive difference metric is proposed to quantify the convergence rate of the trim solutions as more nodes are added. We then compare this gain in accuracy to the simulation time and memory usage by plotting their Pareto fronts. Results reveal that using 6–8 tether nodes (i.e., 7–9 segments) is sufficient. Compared to using 15 nodes, 8 nodes keep the successive difference metric below 2%, run over 13 times faster, and require less than a quarter of memories. Beyond 8 nodes, the differential equation solver appears to enter a new stiffness regime, resulting in rapid rises in computation time and memory usage. These findings suggest that the conservative choices of 10–20 nodes can be relaxed, promising significantly lower computation costs in future studies.