The Static VAR Compensator (SVC) component is based on state variable modeling techniques. It is primarily composed of a thyristor controlled reactor (TCR) and a thyristor switched capacitor (TSC). A state variable formulation solves the differential equations of the system for capacitor voltages and inductor currents. The dependence of the system matrices on the value of the time step is directly proportional, and thus the time step can be readily changed during a run as it is required for the switching elements to be modeled adequately. The Static VAR Compensator model Interfaces to EMTDC as a Norton current source.
The general circuit diagram of the SVC simulated by the Static VAR Compensator component is shown below. All elements shown in this figure are incorporated into the model, and their dynamic behavior is mathematically described within the model by a set of state variable equations.
The TCR elements are connected in delta, and all thyristor switches are modeled as two-state resistances.
The TSC branches are modeled as capacitors. Regardless of the number of TSC branches/stages in operation at a given time, all of these are represented together as an equivalent single capacitor per phase. The value of this equivalent capacitor, and its initial voltage, are adjusted when the TSC switching logic indicates the turning on or off of a capacitor bank. This approach reduces the number of state variables to six for the TSC. Note that such a reduction of the state vector dimensions is possible only because no series inductance is included in the capacitor branch, else each arm would have had to be modeled separately. This is a small price to pay in making the program fast since capacitors are usually switched only a few times in a simulation run.
An Adam's second order closed formula is used for integration, to solve the state variable solution. Since the integration method allows variable time-step with no significant increase of computer CPU time, the turnoff of a thyristor in series with the current carrying inductor (such as in TCR) can be done as close to a current zero-crossing as to provide a minimal voltage spike appearing in the thyristor and inductor voltages. This is implemented by use of smaller time-step when required. When a change of sign of the current value IL is detected, the program does not update the state variable, but re-simulates with sub-multiple time step DELT' = DELT/n. The thyristor switches off close to current zero-crossing and then the program takes one more 'catchup' step to come back into synchronism with the original sampling instant. A value of n between 2 and 5 was found sufficient for most practical cases. The values of DELT and DELT' are under the user control.
