Probabilistic Steady-State Analysis of Power Systems With Photovoltaic Generations

Abstract

Recently, the application of probabilistic methods for power system analyses has become increasingly popular owing to their capability to instill enough confidence in system planner and operator in making more realistic decisions. In the conventional deterministic methods, consideration of a few typically stressed operating conditions are inadequate in solving the present uncertainty problems which are majorly confronted due to the enormous integration of renewable generations along with the conventional load powers. Probabilistic steady-state analysis (PSSA) refers to the adaptation of probabilistic load flow (PLF) to address the aforementioned uncertainties for characterizing the uncertainties in the power system variables referred to as result variables. Among the many renewable sources, photovoltaics (PVs) have experienced a globally increasing significance as its cost per unit is decreasing day by day. PV generation is intermittent and variable with a higher level of uncertainty; their integration to power system greatly affects the power system variables which is a significant concern in the power system studies. Hence, a study focusing on the various aspects of power systems with integration of such renewable resources is the need of the hour. Therefore, this thesis is dedicated towards the PSSA of PV integrated power systems to examine various uncertainty issues that are likely to be combated in transmission systems. The primary requirements for PSSA are mainly of threefold which include uncertainty modeling, power system model development, and application of an uncertainty handling method. This thesis aims at the improvement of each of these facets through suitable modifications and eventually resulting in an elegant PSSA. For the uncertainty modeling, use of the historical record of inputs yields realistic models. For power system expansion and operational planning, such models use the daily time step data corresponding to the time of the year concerning the study of interest. The span of the chosen time series ranges from few months to few years depending on the study requirement or data availability. The daily time series of PSSA inputs such as load iiipower, PV generation, ambient temperature, etc. exhibit complex patterns that are periodic, encompassing predictable components. It is vital to separate such components from the raw data to characterize the unpredictable residuals referred to as preprocessing. In this regard, methods for preprocessing using multiple linear regression is proposed, and are compared with state of the art methods using the data collected from various places in India and USA. The rationale involved in the development of such models is deliberated in detail. Finally, a scenario-based spatiotemporal probabilistic model is developed by adopting the proposed preprocessing, transformation techniques, principal component analysis, and a suitable time series model capable of accurately modeling the trend in the variance of uncertain inputs. Risk-based power system studies considering PV generations facilitate in delimiting the permissible penetration by executing essential steps to hedge systems risks. On this line, a risk assessment of PV arrays integrated to New England 39-bus transmission system is carried out. An improved system model is developed by accounting for the effect of environmental conditions, predominantly, the ambient temperature on the branch parameters by considering the electro-thermal coupling effect. The PLF that embodies the above effect in system model is referred to as temperature-augmented PLF (TPLF). It considers uncertainties in PV generation, aggregate load power, and ambient temperature along with their associated correlations for risk assessment. The effect of increased PV penetrations and variation in TPLF model parameters on the statistics of result variables is analyzed in detail. The expected system over-limit risk indices are calculated and are analyzed for different PV penetrations and input correlations. In general, operational studies require a faster estimation of PSSA. One of the ways to achieve this is through the use of an uncertainty handling method that obtains accurate results in less time. On this line, efforts are made to devise two hybrid methods for PLF and TPLF simulations. Here, \hybrid" refers to the suitable amalgam of two uncertainty handling methods in part or as a whole through suitable modifications. As the thesis focuses on the larger transmission systems, cumulant method is chosen as ivone of the potential methods for hybridization. It is seen that, based on a comprehensive result analysis, the proposed hybrid methods exhibit improved performance in the approximation of multimodal probability distributions of the result variables. For all the above studies, PSSA is carried out on various transmission systems such as New England 39-bus test system, IEEE 14-bus, 57-bus 118-bus test systems, and Indian utility 62-bus test system. MATLAB 7.10 is used to develop the corresponding programming codes for various analyses. Finally, with the aid of the obtained results, the research work in this thesis demonstrates that the proposed models and methods for PSSA are potentially challenging candidates which facilitate in making sensible decisions regarding the planning and operation of PV-integrated power systems.