Abstract:
Modern power system is a complex dynamical system and one of the largest man-made systems. With recent driving forces like environmental concerns over air emissions, the modern power system is evolving towards an even more complex system. So it is necessary to handle the current challenges in power systems with simple approaches and avoid adding further complexity as much as possible. Also the implementation issues should be taken into account to meet the Transmission System Operators' (TSOs') interests. The considered problems in this thesis are related to voltage control and damping control which are two important issues challenging secure power system operation. The first voltage control problem addressed in the thesis occurred during the restoration of the Swedish power system after the blackout in 2003 and is called reactor hunting. Large scale voltage fluctuations are the consequence of the reactor hunting. The common practice used by the Swedish TSO to handle the reactor hunting is to turn off voltage control automatics during the restoration period. That leaves the shunt reactors in manual operation which leads to a longer restoration process. To prevent reactor hunting, an adaptive tolerance band strategy is proposed in the thesis together with two ways to implement it. One is model based and uses short circuit capacity of buses which are going to be energized during the restoration. The short circuit capacity associated to each bus is normally available in the Energy Management System (EMS) in the TSO control center. The second implementation can be completely local and independent of a model. By implementing this strategy, the automatic operation of the reactive shunts will continue during the restoration time, and reactor hunting is eliminated. This should shorten the restoration process. The second voltage control issue addressed in the thesis is related to control of shunt capacitors. Shunt capacitors are commonly controlled using a local scheme, which switches in the capacitor when the voltage at the locally monitored bus is outside a tolerance band. In some cases a shunt capacitor remains unused in a region lacking reactive power just because the local voltage is within the tolerance band. An alternative control strategy proposed in the thesis is called the neighboring scheme. It uses both the local voltage and the voltage at neighboring buses. The neighboring bus voltage is estimated from measurements at the local bus, so this strategy can be implemented locally and communication free which is important for TSOs. In a situation near voltage collapse, this strategy has better performance in the sense of improving the voltage control by connecting more shunt capacitors or connecting them earlier compared to the local scheme. For some scenarios, the voltage collapse that occurs using the local scheme is avoided when using the neighboring scheme. The second actuator used in the thesis for voltage control improvement is VSC-HVDC converters which have the capability to control active and reactive power independently. For emergency voltage control this thesis suggests adjusting active and reactive power set-points to change the AC system power flow. Based on the considered strategy, the active and reactive power set-points are adjusted depending on the disturbance. This control strategy improves the AC system long-term voltage stability and could prevent voltage collapse in some severe scenarios. When designing voltage control systems, the lack of a simple text book size version of NORDIC32 test system for long-term voltage stability study is another issue addressed in the thesis. The NORDIC32 test system is a reduced order model of the Swedish power system but in some cases still a complex test system. In this thesis, we propose the N3area test system which is a text book size version of NORDIC32 with minimum model complexity for our purposes. Applying complex control algorithms to the N3area system and analyzing them is much easier than to the NORDIC32 system. Still it retains a dynamic behaviour quite close to NORDIC32 and reality. The last problem addressed in the thesis is related to inter-area oscillations damping in power systems. These oscillations are becoming a big concern for TSOs since the power systems are getting more and more interconnected. Inter-area oscillations are often limiting the transfer capacity of transmission lines and may even lead to system break up as in the 1996 western North America blackout. Active power modulation is an effective solution to damp out such oscillations. This can be implemented by active power modulation at two points in the network, using for example VSC-HVDC links. Also single-point active power modulation using actuators like Energy Storage (ES) works well. Singlepoint reactive power modulation using actuators like SVC indirectly controls the active power and is also efficient. Proportional control of active power with local frequency as input is used in reality today for HVDC links. This type of damping controller can be applied for the ES and can also be translated for SVC damping controller. Implementing such proportional damping controllers is simple as they use local feedback signals. However, the damping of the inter-area mode is limited due to nearby zeros, which is evident in the associated root locus plot. It is therefore important to use the optimum gain to achieve the maximum possible damping. Gain selection is normally done using visual inspection of the root locus or through optimization. In this thesis, we propose the impedance matching based gain selection for the VSC-HVDC, ES and SVC damping controllers. It gives a physically based criterion for the optimum gain selection to reach the maximum possible damping of the mode with the greatest mode observability and controllability which depends on the actuator location while not affecting negatively the other modes in the system. The proposed approach may be used as basis for a controller that is self-tuning which is an important feature since the power system operating points are changing a lot. Also it is simpler for implementation in reality compared to the root locus inspection or application of advanced optimization methods for gain selection.

Dowload thesis here (four papers which are currently under review are not included):

http://www.iea.lth.se/publications/Theses/LTH-IEA-1086.pdf

Supervisors:
Prof. Olof Samuelsson (Lund University, Sweden)
Assoc. Prof. Jörgen Svensson (Lund University, Sweden)

External examiner:
Prof. Kjetil Uhlen (University of Science and Technology in Trondheim, Norway)

Examination committee:
Prof. Massimo Bongiorno (Chalmers, Sweden)
Prof. Mehrdad Ghandhari (Royal Institute of Technology in Stockholm, Sweden)
Prof. Claus Leth Bak (Aalborg University, Denmark)
Dr Thomas Smed (Vattenfall, Forsmark, Sweden) (alternate)

Chairman:
Associate Prof. Ulf Jeppsson (Lund University, Sweden)