Ion thrusters achieve very high specific impulse by accelerating charged particles across a potential difference using electrostatic force fields. These force fields are set up by two or three metal grids with thousands of small apertures, called the ion optics. Work currently being done in this area includes the modeling and prediction of the erosion of the ion optics due to impacting ions. The current ion engine of interest is NASA's NEXT thruster, a higher power version of the very successful NSTAR thruster.
The NEXT ion thruster. Images courtesy of NASA GRC
The device and plume flows are modeled using a combination of the direct simulation Monte Carlo (DSMC) and Particle-In-Cell (PIC) techniques. These methods simulate the behavior of uncharged and charged particles respectively. The ultimate goal of this work is to develop computational models to predict operational characteristics of ion thrusters to aid in new engine design. The output of the device model can then be used to compute the plume to assess spacecraft integration issues. The model simulates a single axisymmetric aperture in the thruster grids.
Currently, ion thruster modeling is being carried out by Jerry Emhoff, a fifth year graduate student. He is working with the NASA Glenn Research Center on predictions for ion optics erosion in the NEXT thruster. Erosion of the optics is the primary failure mode of an ion thruster, and is at the same time very difficult to measure experimentally. The combination of these aspects make computational simulation of the erosion a priority in modeling research.
Most recently, a study of the accuracy of the potential field solver used in the model was performed. This study showed that the potential solver mesh needed to be more highly refined in order to obtain accurate results. Figure 1 shows the decrease in error as the mesh is refined.
Figure 1. Decrease in potential field error as the mesh is refined.
The need for a higher level of refinement in the simulations also leads to a longer computational running time. To help decrease the running time, a multigrid potential field solver was developed to replace the Alternating-Direction Implicit (ADI) solver used previously. Figure 2 shows the increase in speed produced by the multigrid solver.
Figure 2. Computational time as a function of the residual error level.
Concurrently with the NEXT thruster modeling, Jerry is also developing a mesh-less ion optics simulation. The potential solver in this case is a treecode, which groups particles in the domain and computes forces based on the group properties. The advantages of the meshless approach are that the grid geometry can be simulated exactly without the use of cut-cells or other approximations, and the particle forces are computed more accurately. It may also be possible to reduce the computation time needed by the simulation. In Figure 3, three plots of the potential field are shown. The first uses the treecode solver with cusped ion optics, where the cups are represented exactly. The second uses the treecode solver without cusps, and the third uses the traditional gridded solver.
Figure 3. Comparison of the different field solvers.
AcknowledgmentsFunding for this work provided in part by NASA Glenn Research Center.
Emhoff, J. W. and Boyd, I. D., Progress In NEXT Ion Optics Modeling, AIAA-2004-3786, Presented at the 40th AIAA/ASME/SAE/ASEE Joint Propulsion Conference, July 2004, Fort Lauderdale, FL.
Emhoff, J. W. and Boyd, I. D., Grid Erosion Modeling of the NEXT Ion Thruster Optics, AIAA-2003-4868, Presented at the 39th AIAA/ASME/SAE/ASEE Joint Propulsion Conference, July 2003, Huntsville, AL.
Emhoff, J. W., Boyd, I. D. and Shepard, S. P., Numerical Analysis of NEXT Ion Thruster Optics, IEPC-03-110, Presented at the 28th International Electric Propulsion Conference, March 2003, Toulouse, France.