Development of the 3D PIC-DSMC code Pantera and its application to Air-Breathing Electric Propulsion

Spacecraft orbiting the Earth at altitudes below 250 km experience significant drag from the rarefied atmosphere, causing them to de-orbit in a matter of days if this is not compensated. A concept called Air-Breathing Electric Propulsion (ABEP) proposes to collect the atmospheric gas through an intake and use it as the propellant in an electric thruster to compensate the drag. This removes the burden of carrying propellant at launch, but introduces additional requirements and challenges.

Ground testing is fundamental in the design and optimization of an ABEP system. The DRAG-ON facility at VKI was designed and built to test the performance of intakes for ABEP. However, faithful reproduction of the orbital flow is practically unattainable. Therefore, numerical simulation is required to explain the experimental measurements and to extrapolate the tests in laboratory conditions to the performance in orbit.

With this goal in mind, we developed Pantera: a Particle-in-Cell code with collisions computed through the Direct Simulation Monte Carlo or the Monte Carlo collisions method. The use of an unstructured discretization allows us to tackle the simulation of geometries of engineering interest. The main obstacle to the simulation of plasmas, however, remains the presence of length and time scales of vastly different magnitude. To alleviate this problem, we formulated and implemented semi- and fully-implicit PIC temporal integration schemes which conserve energy exactly. These are stable even when the smallest scales are under-resolved, allowing to lower the computational cost. We will give an overview and comparison of such methods and examples of their application. Gas-surface interaction plays an important role in ABEP. We will present the washboard model, which we found necessary to correctly reproduce the experiments in the DRAG-ON facility. Finally, we will present recent work on code-to-code verification, where we used error quantification techniques adapted to stochastic methods, and discuss possible future directions.