SVC dynamic analytical model
Summary (3 min read)
Introduction
- The model consists of three individual subsystem models: an AC system, a SVC model and a controller model, linked together through d-q transformation.
- The model derivation for a different system will be similarly tedious and the final model form is not convenient for the application of standrd stability studies and controller design theories.
- It also seeks to offer complete closed loop model verification in the time and frequency domains.
- The AC system model is also based on [11], with the introduction of an additional local load and a variation in system impedance to represent different and extreme system strengths.
A. Model structure
- To avoid pitfalls with modeling complex systems, the system model is here divided into three subsystems: an AC system model, a SVC model and a controller model.
- Each subsystem is developed as a standalone state-space model, linking with the remaining two subsystems and with the outside sgnals.
- The variables with subscript “out” are the outside inputs and outputs.
- The input matrices, Bij, take the second index “j” from the particular input-side connecting subsystem (i.e.: Bacco is the AC model input matrix that takes input signals from the controller), and the output matrices Cik have the second index “k” associated with the linking subsystem that takes the particular output vector.
B. AC System Model
- The AC system model is linear, developed in the manner described in [8] and [9], and only a derivation summary is presented here.
- A single-phase dynamic model is developed firstly, using the instantaneous circuit variables as the states.
- The test system uses a third order model with iL1, iL2 and v1 as the states.
- A phase “a” model is given below (to increase clarity of presentation the authors consider only one input link, one output link and only one D matrix): acacoacacocoacaacacoacaco acacoacacoacaacaaca uDxCy uBxAx += += o (2) where the subscript ”aca” denotes phase a of the AC system.
- To enable a wider frequency range dynamic analysis and coupling with the static coordinate frame, the above model is converted to the d-q static frame using Park’s transformation [8],[12].
C. Static VAR Compensator Model
- The static VAR compensator under consideration is a twelve pulse system with two six pulse groups in ∆ connection and coupled with the network through a single, three-winding transformer with Y and ∆ secondaries [10],[11].
- This equation cannot be directly linearised since the SVC model is developed in the AC frame with oscillating variables, (i.e. )cos( ϕω += tVv 22 ) whereas the firing angle signal is derived as a signal in the controlle reference frame (i.e. a non-oscillating signal).
- Equation (14) is in the AC coordinate frame, and the following term: )cos( osvc o svc o tKVKv ϕωφφ +∆=∆ 22 (15) is an artificial oscillating variable that has a varying magnitude and a constant angle equal to the voltage nominal angle.
- Subsequently, using the d-q components of the inputs and outputs, this model is linked with the other model units.
- It should be noted that the transformer impedance (Lt) must be included in this model since the eigenvalue analysis proves that this parameter has noticeable effects on system dynamics.
D. Controller Model
- The controller model consists of a second order feedback filter, PI controller, Phase Locked Loop (PLL) model and transport delay model, as shown in Figure 3.
- The PLL system is of the d-q-z type and its functional diagram is given in [14] and [10], whereas the state space linearised second-order model is developed in [8].
- The delay filter does not have dynamic equivalent in the actual system.
- This simplified continuous-element modeling of a discrete phenomenon has limited accuracy, but the model application value is much increased with the continuous form and, as demonstrated in the following sections, accuracy proves satisfactory for most applications.
- During the proposed model verification, simulation studies have suggested that the value of approximately Td=2.85ms is used, which is in agreement with the above rec-.
E. Model Connections
- The above three models are linked to form a single system model in the state-space form.
- The final model has t e following structure: outsssout outssss uDxCy uBxAx += += o (16) where “s” labels the overall system and the model matrices are: D matrices are assumed zero in (17) since they are zero in the actual model and this noticeably simplifies development.
- As has the subsystem matrices on the main diagonal, with the other sub-matrices representing interactions between subsystems.
- The model in this form has advantages in flexibility since, as an example, if the SVC is connected to a more complex AC system only the Aac matrix and the corresponding input and output matrices need modifications.
B. Frequency domain
- The two test systems were also tested against PSCAD in the frequency domain.
- It is seen that the PLL gains have significant influence on the system dynamics and that the frequency of the dominant oscillatory mode reduces, accompanied by a small reduction in mode damping (branch “a”).
- Model verification in the time and frequency domains against a PSCAD simulation confirmed very high accuracy for f<25Hz, and fair accuracy even beyond the first harmonic frequencies.
- A. Ghosh, G.Ledwich, “Modelling and control of thyristor controlled series compensators” IEE Proc. Generation Transmission and Distribution, Vol. 142, No 3, May 1995.
- He worked as a post doctoral fellow at University of Canterbury, New Zealand during 1992-1993 and then joined Trans Power New Zealand Limited where he is currently employed as aSenior Network Support engineer.
Did you find this useful? Give us your feedback
Citations
215 citations
128 citations
76 citations
73 citations
55 citations
References
13,467 citations
"SVC dynamic analytical model" refers methods in this paper
...To enable a wider frequency range dynamic analysis and coupling with the static coordinate frame, the above model is converted to the static frame using Park’s transformation [8], [12]....
[...]
183 citations
"SVC dynamic analytical model" refers background in this paper
...The modeling principle reported in [6] employs rotating vectors that are difficult to use with stability studies, and only considers the open loop configuration....
[...]
107 citations
"SVC dynamic analytical model" refers methods in this paper
...The model presented in [2] uses a special form...
[...]
102 citations
"SVC dynamic analytical model" refers background in this paper
...The importance of having a state-space represented, linear continuous system model, is well recognized in [4]....
[...]
92 citations
Related Papers (5)
Frequently Asked Questions (11)
Q2. What is the common use of a PLL?
A PLL is typically used with thyristor converters to provide the reference signal that follows the synchronizing line voltage or current.
Q3. What is the effect of the PLL on the system dynamics?
It is seen that the PLL gains have significant influence on the system dynamics and that the frequency of the dominant oscillatory mode reduces, accompanied by a small reduction in mode damping (branch “a”).
Q4. What is the d-q component of the rotating susceptance?
In order to link the d-q components of the rotating susceptance (15) with the controller module, these components are further converted to magnitudeangle components using the x-y to polar co-ordinate transformation [8].
Q5. What is the way to verify the accuracy of a PLL?
Model verification in the time and frequency domains against a PSCAD simulation confirmed very high accuracy for f<25Hz, and fair accuracy even beyond the first harmonic frequencies.
Q6. What is the corresponding input-side index of the controller?
The input matrices, Bij, take the second index “j” from the particular input-side connecting subsystem (i.e.: Bacco is the AC model input matrix that takes input signals from the controller), and the output matrices Cik have the second index “k” associated with the linking subsystem that takes the particular output vector.
Q7. What is the way to describe the phase response in MATLAB?
Regarding the overall time and frequency domain responses, and being aware of the phase response errors, it can be concluded that the model has reasonably good accuracy when employed as a design and analysis tool for phenomena such as subsynchronous resonance, or interactions with other fast FACTS/HVDC controls.
Q8. What is the effect of the PLL on the speed of response?
Although not shown in Figure 10, the increase in the PLL gains increases the speed of response and it is suggested that this effect on the positioning of the dominant mode can be exploited in the design stage to improve performance, or to avoid negative interactions at a particular frequency.
Q9. What is the test system in use?
The test system in use consists of a SVC connected to an AC system that is represented by an equivalent impedance and a local load, as shown in Figure 1.
Q10. What is the susceptance value in the steady state?
Assuming small perturbations around the steady state the authors have:)( 222 vvv o ∆+= (9) )/(// tcrtcrtcr LLL 111 0 ∆+= . (10) Small perturbations are justified assuming an effective voltage control at the nominal value.
Q11. How many times reduced gains are seen in Figure 10?
As the gains are reduced, the eigenvalues migrate from the original “x” to the location “o”, representing ten times reduced gains.