Hello,
I am currently working with attiude propagation. Following what written in Re: [Orekit Users] Orekit attitude numerical integration, it seems to me that Orekit propagates the attitude based only on its imposed kinematic law regardless of the force models introduced in the propagator. For example, supposing a force acting unbalanced on a satellite, is it correct to say that this force generates an acceleration which affects only the orbital propagation whilst the attitude is completly unaffected by the generated torque and therefore it propagates as if the force was not present at all (even if physically it is not true)?
Yes, this is true.
The orbit is propagated using the forces considered to be acting on the center of gravity, and the attitude law is applied later.
The attitude laws provided in the library by itself do not perform any additional dynamics computation, they just apply a “perfect” kinematics model. It is theoretically possible to create an AttitudeProvider that would take torques into account and perform its own integration, but this has to be done by users. They must be aware however that due to the way numerical integrators, step handlers and events handler work together, the attitude provider may see calls at apparently random times. For example you could have the attitude provider called at t, t+1s, t+2s, t+3s, t+4s, and then t+0.1s, t+0.12345s, and then again t+4s. If an integrator is used, it may get lost as time rewinds.