what is DSTATE?

Thank you for the answer. The was the thing that came to my mind after RJG@PSM's answer. I think I'll go with this. I think what you suggest will work fine in quasi steady-state operation. During the transient, I think this is the best I can get. Thank you again for the answer.

Thank you for the answer. I am changing the state variable in every timestep. In this case, how do we know the derivative of the state variable? I mean we don't know the diff. algebraic equations corresponding the state variables. Is it OK if we just write something other than 0?

Ideally, as mentioned above, DSTATE=0 at the start of your simulation will ensure that you get a "flat start" or "steady-state initial conditions". After that, you need to calculate DSTATE at each time step to do the numerical integration.

@jfconroy Do I have to provide a derivative to PSS/E? Doesn't it have Differential Algebraic Equations which it can make a transition to next state(k+1)?

If you are writing your own dynamic models, then you have to write the equations for DSTATE for each state variable that you need. If you are using "built-in" dynamic models, then there is no need for you to define DSTATE for any state variable.

If you are seeing initialization warnings about DSTATE, when you are using "built-in" dynamic models, that means you need to adjust some load flow or dynamic model parameters, you do not need to re-write equations.