A typical problem with thermodynamics is the misunderstanding of its meaning.
Thermodynamics describes the behavior of energy in a system and its subsystems. It is pretty abstract and subjective.
- Subjective, because it deals only with one closed system, e.g. the container (real systems are mostly open, and have no boundaries, think on a tree, a cow or the sea) and only one level of subsystem: particles (real subsystems can process energy for multiple levels, e.g. a particle can be split into atoms, ergo allowing a larger entropy on the same system).
- Abstract, because it simplifies energy to be manifested in elementary forms: heat, pressure, etc. In reality, energy has multiple forms, not just those addressed by thermodynamics. For example, electromagnetic, atomic, gravitational, chemical, etc. If you need to include chemical energy in your thermodynamic assessment of a system, see chemical thermodynamics, and so on.
So, under such constraints, thermodynamics says: a) energy conserves (even if internal gravitational energy is present) b) energy tends to disperse within the subsystems of a system (including gravitational). This means that all the subsystems will tend to have the same energy. That is the second law's significance.
If you think it, the first law describes the SYSTEM, the second law describes the SUBSYSTEM, the zeroth law describes our subjectivity (just think on it: temperature is just a feeling, the zeroth law formalizes it), and the third law just describes the limit of the second law. I would personally state that systems are fractals: they are made by systems and they contain systems, infinitely. But not for thermodynamics, there are only two systems in thermodynamics, the container and the particle; the third law defines the limits of the subsystem (where no more subsystems are possible), by determining the point where entropy is zero; the zeroth law defines the limits of the system (where no more supra-systems exist, by defining what is our understanding of energy (the definition of temperature), in fact, some cientists have proposed to express temperature in joules.
When you state that particles tend to be localized in a zone of the container, you are implying a subjective interpretation of the behavior of the molecules. Within classical thermodynamics, molecules will move and distribute homogeneously (you can understand that as disorder, and I can interpret that as order: order is a subjective appreciation). But in your thermo-gravitational-dynamics, molecules might fall down (implying that they are not perfectly elastic, as classical thermodynamics do). Perhaps they will. So you are concluding that the second law is failing and entropy decreases in a closed system, because we can clearly see that molecules tend to order.
But that is wrong! The second law is not about disorder! It just states that energy will disperse across the subsystems of the system. Such is a common misunderstanding. Entropy is not disorder, but moreover energy dispersal (do not forget: there are only two types of systems in thermodynamics: containers and particles; entropy means energy dispersal within particles; just that). So even in the case that particles group inside the container, entropy reaches its maximum (within the SUBSYSTEMS) , spontaneously, energy conserves (within the SYSTEM), and order or disorder are subjective appreciations (within our OPINIONS).