I think you are going to be better off if you add your stan model in separate post. Or even better you can try to break it down in a smaller model where you have the same problem. Ideally, with complicated model you want to build up from small to big model. Once you hit a stumbling block than you can tell which step is causing the problem.
As a guess now, I would look for two opposite things. Maybe the model is not identified. Different combinations of parameters give you almost identical posterior. This can be because the mathematical model is poorly identified, it could also stem from a coding error. For instance, a loop that is not correctly implemented.
The opposite problem is that the model is a deterministic representation of the simulated data. If the correct model can perfectly predict the outcomes in the simulated data, the posterior has probability 0 for any non correct model and 1 for the correct model. Stan doesn’t handle does types of discontinuities well. (The one chain that gets stuck at the correct value could be an indication of something like this).