Messages

MIL-101 is very large indeed. RASPA is not optimized for speed.  You can use a debugger to see where it goes wrong.

9% at those high pressure is still very, very good I would say. The validity of such a simple methane model is not "endless". Any classical model will show weaknesses at extreme conditions. Perhaps a full-atom model might do better, but these are harder to paremeterize.

1) As many cycles and molecules as you need. The more the better, but the slower the simulations. That's always a trade-off.
In equilibrium, the macroscopic properties no longer change. The amount of cycles needed to reach equilibrium depends on the system.
To limit finite size effects you need a large system. In practice you can simulate various system sizes as see when your properties converge to a plateau value as a function of system size.

2) in a NVT or muVT or Gibbs NVT simulation the energy fluctuates. This energy fluctuation is related to a physical property called the heat capacity.

3) There are no rules of thumbs, but rather it will take experience to get a feel for this.

Yes, all Qeq are differently implemented. The code in RASPA is based on the method of Wilmer and Snurr.
https://pubs.acs.org/doi/10.1021/jz3008485

The 'pseudo_atoms.def' contains predefined atoms, and this is often the way to go for molecules.
This especially important for united atom types etc.

For frameworks, read from a CIF file, there is always an element defined (_atom_site_type_symbol), and the type (_atom_site_label) is automatically added when found. The mass is a lookup from a database in 'src/scattering_factors.c'.

This function writes out a VTK-file, a 3D histograms of the density. See the examples on Visualization.
You could also read this file yourself.

The DREIDING force field uses specific functional forms and parameters. You can look these up and use (most of) these in RASPA.

Input looks, at first sight, okay to me. Remove as much as possible, i.e. component 1-3 etc, and check if it works. Then from add the components one by one to see where it goes wrong.
In general, change stuff in small steps, otherwise it is harder to debug.

I don't think this is currently possible. The implementation of the CBMC algorithm used for the insertion is the limitation here.

No, the pressure is indeed not used in this example.
I will remove the pressure line in the example.

Have a look at the example rigid molecules in RASPA, e.g. CO2.
Numbering starts at zero.

That is hard to say based on this information. Always try with the latest version (2.0.45).
If it crashed, then use a debugger to find out why it crashes (section 5.5 in the manual).

No. For MD, you're better off using LAMMPS which is parallel and very fast.

Adsorption is most difficult to compute at low temperatures where the system might get stuck in meta-stable states. With parallel-tempering and hyper-parallel tempering you overcome these free energy bottle necks. Also techniques like expanded ensembles help to overcome the problem of low insertion/deletion rates.

 I am not using windows myself, but you just have to setup your setup the compilers.
There are also unix-like environments under windows like the 'WSL' (windows subsystem for linux) and MSYS.