Making the Most out of Distributed Generation without Endangering Normal Operation: A ModelBased TechnicalPolicy Approach
 Publication Year:
 2012

 Bepress 1542

 Bepress 104
 Repository URL:
 http://repository.cmu.edu/dissertations/141
 Author(s):
 Tags:
 Electrical and Computer Engineering
thesis / dissertation description
In this dissertation we introduce a modelbased approach for efficiently locating and operating distributed generation (DG) without endangering stable system operation. The proposed approach supports quantifiable policy making based on technical design. The model used is structural and it comprises local models of DGs and loads interconnected via distribution grid system. While similar model structure can be used to represent meshed transmission grids, identifying model properties unique to distribution systems sets the basis for interpreting power delivery losses as the key measure of the overall system efficiency. It furthermore sets the basis for designing decentralized control specifications necessary to ensure systemwide stability. Once the underpinnings of the technical design are understood, the findings are used to propose a modelbased quantifiable policy design to support process of integrating and operating DGs in distribution systems.We first investigate efficient integration of distributed generation on the distribution side of electric energy systems. We introduce a notion of efficiency in distribution systems which is uniquely determined by the fact that DG units are clean and inexpensive; because of this DGs are always scheduled and there is no need for economic dispatch. This points to the fact that the main measure of efficiency is loss minimization. This notion helps us in streamlining specific methods for optimizing losses both in planning and operation. At the planning stage the best location is found, and in operation optimal voltage dispatch is done to reduce losses. We show that a 10% penetration of DG units can reduce up to 50% of distribution losses, if DG units are strategically located and optimally operated in distribution systems.One possible problem with optimal placement of DG units may be an overly high sensitivity of their response to even small perturbations from normal conditions. Therefore, a very efficient distribution system with optimallyplaced DG units may not be robust in operations. In order to assess robustness of distribution energy systems with respect to small disturbances, we model distribution systems as dynamical systems. We show that because of the strongly coupled voltage/realpower interdependencies in power flows of distribution systems, it is no longer possible to use a decoupled realpower dynamic model which neglects the effects of voltage dynamics. This conclusion is a direct consequence of a nonnegligible resistancereactance ratio in distribution systems which differentiate them from the typical transmission systems. Therefore, only coupled models should be used for stability analysis and for control tuning of DGs in distribution systems.Using such a dynamic model we show that distribution systems with high penetration of DG units can exhibit frequency and/or voltageinstabilities when power plants have conventionally tuned control. Such instabilities are particularly pronounced when the DG units are electrically close. Gerschgorin circle theorem and participation factorbased methods are used to identify the main cause of instabilities as being the interactions of the local DG dynamics through the distribution power grid. Since the proposed dynamical model structure allows us to represent any type of DG plant and its local control, stability analysis can be performed for a general type of a DG using these methods to determine bounds on interactions between each specific DG and the rest of the system so that no interactions occur.These bounds are dependent on the machine type and parameters, the local control and the grid parameters. Some DGs may not have sufficient control as measured in terms of these bounds, and, these are the ones which require enhanced control to ensure systemlevel stability without unstable interactions, as discussed next.The severity of dynamical problems in specific distribution systems with DGs depends on the technology and control of DGs and on the electrical distances between the DGs. Typical DGs are either synchronous machines or induction machines whose inertia may be much smaller than the inertia of large generators. Their local control may range from no control, through wellunderstood governorexcitation control of synchronous machines, through power electronically controlled inverters of synchronous and/or induction machine type DGs (power system stabilizers (PSS) and/or doubly fed induction machines (DFIG)).In this dissertation we have studied stability problems in systems with DGs being small and/or medium size synchronous machines controlled by governorexcitation systems and/or by pitch control combined with PSS control. We assess possible instabilities in such systems when controllers are tuned on a standalone machine connected to the impedance representing the rest of the system (today's practice). We show that a more systematic fully decentralized, and, therefore, simple, control design proposed, in this dissertation, could stabilize synchronous machinetype DGs, such as diesel and hydro plants, without inverter control. Moreover, synchronous machinetype wind power plants can be stabilized in a decentralized way by combining advanced pitch control and/or PSS control.Based on the above technical findings we propose a policymaking process for giving guidelines: (1) to best locate candidate DG units; (2) to dispatch set points on the voltage controllers of DGs in coordination with dispatching set points of other voltagecontrollable equipment for ensuring minimal losses in operations; and, (3) to enhance the existing control of the DGs and/or deploy new enhanced decentralized control. Because the solutions are systemdependent, simple onesizefitsall policies are no longer viable; instead, policy decisions must be supported by software for placing the DGs and for designing their voltage dispatch and control. This approach leads to systematic institutional agreements and policies needed to support large penetration of DG units while ensuring both efficiency and robustness of distribution energy systems.