Question about modification of PIController

[xiaoheimai95]

Dear Dr. Cui and Jinning,
I modified the model PIController in order to add an external parameter utd and utq to inffect the voltage of VSG. The main modification of PIController is as follows:

class PIControllerwithut(Block):
    """
    Proportional Integral Controller with external control variable ut.
    """

    def __init__(self, u, kp, ki, ut, ref=0.0, x0=0.0, name=None, tex_name=None, info=None):
        Block.__init__(self, name=name, tex_name=tex_name, info=info)

        self.u = dummify(u)
        self.kp = dummify(kp)
        self.ki = dummify(ki)
        self.ut = dummify(ut)
        self.ref = dummify(ref)
        self.x0 = dummify(x0)

        self.xi = State(info="Integrator output", tex_name='xi')
        self.y = Algeb(info="PI output", tex_name='y')

        self.vars = OrderedDict([('xi', self.xi), ('y', self.y)])

    def define(self):
        r"""
        Define equations for the PI Controller with external control variable ut.

        Notes
        -----
        One state variable ``xi`` and one algebraic variable ``y`` are added.

        Equations implemented are

        .. math ::
            \dot{x_i} &= k_i * (u + ut - ref) \\
            y &= x_i + k_p * (u + ut - ref)
        """

        self.xi.v_str = f'{self.x0.name}'
        self.xi.e_str = f'{self.ki.name} * ({self.u.name} + {self.ut.name} - {self.ref.name})'

        self.y.v_str = f'{self.kp.name} * ({self.u.name} + {self.ut.name} - {self.ref.name}) + {self.x0.name}'
        self.y.e_str = f'{self.kp.name} * ({self.u.name} + {self.ut.name} - {self.ref.name}) + ' \
                       f'{self.name}_xi - {self.name}_y'

However, when I test the effect of the modification, it came up with the problem of initialization failure on the case of SMIB.txt, I will upload the modified SMIB.txt and init_info.txt later.
So is it possible for you to help me figure out this problem?
info.txt
SMIB.xlsx

[jinningwang]

Check the v_str and e_str of xi. As discussed in documentation, v_str is used to initialize the equation.
Please ensure the v_str and e_str are consistent.

[jinningwang]

It is a little bit weird to me that the Eqn. mismatch seems to be 0. Is it the numerical issues that they are not actually 0?

The documentation discussed the value and equation strings: Development - Variables

Further, in the Example - TGOV1, there is a definition of algebraic variable pd.

  self.pd = Algeb(info='Droop output',
              v_str='tm0',
              e_str='(wd+pref)/R-pd')

Given its v_str and e_str, it should satisfy (wd+pref)/R-tm0=0, otherwise TDS.init() will fail.

Did this answer your question?

[xiaoheimai95]

When I change the equation of Id, Iq, and the logic of PIcontroller, the problem has been solved, however, it comes out with a new initialization problem.
The code of Id and Iq is as follows:

self.Id = Algeb(tex_name='I_d',
                info='d-axis current',
                v_str='Id0 + Kpvd * (vd + utd - vref2)',
                diag_eps=True,
                )
self.Iq = Algeb(tex_name='I_q',
                info='q-axis current',
                v_str='Iq0 + Kpvq * (vq + utq)',
                diag_eps=True,
                )

The code of the new PIcontroller:

self.xi.v_str = f'{self.x0.name}'
self.xi.e_str = f'{self.ki.name} * ({self.u.name} - {self.ref.name})'

self.y.v_str = f'{self.kp.name} * ({self.u.name} + {self.ut.name} - {self.ref.name}) + {self.x0.name}'
self.y.e_str = f'{self.kp.name} * ({self.u.name}  + {self.ut.name} - {self.ref.name}) + ' \
               f'{self.name}_xi - {self.name}_y'

And the problem is:

Suspect initialization issue! Simulation may crash!

    Name     | Var. Value | Eqn. Mismatch
-------------+------------+--------------
 Pe REGCV3 5 | 0.350      | 0.154
 Qe REGCV3 5 | 0.169      | -0.154

[xiaoheimai95]

The reason for this modification is that I only intend to add the control variable ut to the PI output part, then Id and Iq will also change because of the change of PIvd_y and PIvq_y here, not purely the original set initial values of Id0 and Iq0.
My guess is that the v_str and e_str of Pe and Qe don't match anymore because of the change in Id and Iq, and I've observed the definitions of these two variables:

        self.Pe = Algeb(tex_name='P_e',
                        info='active power injection from VSC',
                        e_str='vd * Id + vq * Iq - Pe',
                        v_str='Pref')
        self.Qe = Algeb(tex_name='Q_e',
                        info='reactive power injection from VSC',
                        e_str='- vd * Iq + vq * Id - Qe',
                        v_str='Qref')

However the initial values of Pref and Qref I don't think should then be changed either. But I'm not so sure about my guesses and would like to see your opinion on it

[xiaoheimai95]

Is it possible to ensure consistency between v_str and e_str for Pe and Qe by changing the v_str in the code to the following format?

self.Pe.v_str = 'vd * (Id0 + Kpvd * (vd + utd - vref2)) + vq * (Iq0 + Kpvq * (vq + utq))'
self.Qe.v_str = '-vd * (Iq0 + Kpvq * (vq + utq)) + vq * (Id0 + Kpvd * (vd + utd - vref2))'

[xiaoheimai95]

Found out that this method will influence the value of dw, a, and v:

Suspect initialization issue! Simulation may crash!

    Name     | Var. Value | Eqn. Mismatch
-------------+------------+--------------
 dw REGCV3 5 | 0          | -0.154
 a Bus 8     | -0.022     | -0.154
 v Bus 8     | 1.030      | 0.155

[jinningwang]

Following comments for your consideration:

1. As the block PIControllerwithut is newly developed, we could focus on it firstly if initialization failed.
2. Since you developed a new block, two possible issues might occur: 1) the block definition itself has initialization issue, 2) the models that have this block has inappropriate initialization values.
3. 2.1 can be inspected by comparing with other existing block definitions, and their occurrence in models.
4. Once 2.1 is checked, 2.2 can be checked in the involved model.

As developing a new block can be a little bit complex, here is another easier way. You can skip developing a new block but implement the function you need in a model directly. It should be easier for initialization debug.

[xiaoheimai95]

As you mentioned, I modified the code of VSGOuterPIModel part as follows:

class VSGOuterPIModel:
    """
    Outer PI controllers for REGCV4 with external control variable ut.
    """

    def __init__(self, vderr: str = 'vd-vref2', vqerr: str = 'vq'):
        # PI controller for d-axis
        self.xid = State(info="d-axis PI controller integrator",
                         tex_name='x_{id}',
                         v_str='Id0 - Kpvd * (vd0 - vref2)',
                         e_str=f'Kivd * ({vderr}) * Qlim_zi')
        self.Idref = Algeb(info="d-axis current reference",
                           tex_name='I_{dref}',
                           v_str='Id0',
                           e_str=f'Kpvd * ({vderr} * Qlim_zi + xid + utd) - Idref')

        # PI controller for q-axis
        self.xiq = State(info="q-axis PI controller integrator",
                         tex_name='x_{iq}',
                         v_str='Iq0 - Kpvq * vq0',
                         e_str=f'Kivq * ({vqerr})')
        self.Iqref = Algeb(info="q-axis current reference",
                           tex_name='I_{qref}',
                           v_str='Iq0',
                           e_str=f'Kpvq * ({vqerr} + xiq + utq) - Iqref')

However, when I run TDS.init(), still got the result of Initialization failed:
Suspect initialization issue! Simulation may crash!

     Name       | Var. Value | Eqn. Mismatch
----------------+------------+--------------
 Idref REGCV4 3 | 0.396      | 0.100        
 Idref REGCV4 5 | 0.340      | 0.100        
 Iqref REGCV4 3 | -0.154     | 0.100        
 Iqref REGCV4 5 | -0.164     | 0.100        


Calculated TDS.h = 0.0333333
Initialization for dynamics completed in 0.0482 seconds.
Initialization failed!!

Also, because I wanted to impose some limitations on Qe, I added the Qlim_zi flag, which at this point has a value of 1 and doesn't affect what we're talking about.