Extremely-Low Lunar Orbits (eLLOs; lunar altitudes <= 50 km) offer unique scientific and commercial opportunities by bringing spacecraft into very close proximity to the lunar surface. Unfortunately, at these altitudes, the highly non‐spherical lunar gravity field coupled with the slow lunar rotation rate induces rapid eccentricity growth that, if left unchecked, will typically drive a spacecraft to impact within weeks. Classical ‘frozen’ orbits, which avoid long-term drifts in orbital elements, do not exist at these altitudes, so stationkeeping delta-V costs have historically been significant for many missions.

Recent studies have revealed that, under these perturbations, the eccentricity vector in eLLOs interestingly undergoes a near-periodic evolution largely independent of its initial location. By applying a simple translation in the eccentricity plane—rather than full orbit propagation with high-degree spherical harmonics—one can predict the subsequent evolution on the eccentricity vector plane with low computational cost. Dubbed the ‘translation theorem’, this approach has seen use on missions such as LRO and GRAIL. However, this process has generally required manual input from experts.

In prior studies by the main author, a greedy algorithm was used to automatically select optimum translation targets for an impulsive spacecraft by maximizing the interval until stationkeeping limits are violated. In this work, we extend the framework to solar sailing, which is more difficult to control because they exhibit almost continuous thrust and are dynamially underactuated. Furthermore, because of the low altitudes, the inclusion of lunar albedo and eclipse shadowing are essential for realistic dynamics. For this reason, the greedy stationkeeping strategy is modified to only consider the plausible directions for control. Then, we formulate a lightweight Sequential Convex Programming (SCP) routine that refines these greedy maneuvers into smooth, feasible control profiles for a solar sailing spacecraft. The dominant computational bottleneck remains the evaluation of the high-degree spherical harmonic model.