Traženje optimalnoga razmještaja distribuirane proizvodnje u srednjenaponskim elektroenergetskim mrežama predstavlja nelinearan kombinatorički optimizacijski problem koji se u znanstvenoj literaturi javlja početkom 90-tih godina prošloga stoljeća. Ovisno o funkciji cilja koju se optimizira, prisutan je niz interpretacija optimizacijskoga problema. Također je do sada razvijen i čitav set metoda za njegovo rješavanje, uključujući egzaktne i heurističke metode. U doktorskoj disertaciji se pristupilo formulaciji takve funkcije cilja koja će istovremeno zadovoljiti interese investitora distribuiranih izvora i interese Operatora distribucijskog sustava. Cilj je, procjenom investicijskih projekata svih distribuiranih izvora i procjenom ušteda Operatora distribucijskog sustava, pronaći optimalan razmještaj distribuiranih izvora u srednjenaponskoj mreži. Potonje podrazumijeva odabir njihove optimalne snage i mjesta priključenja. Pritom u obzir moraju biti uzeta naponska i strujna ograničenja mreže, godišnji rast tereta te moguća ograničenja na lokaciju i priključnu snagu svakog distribuiranog izvora. Istraživanje je usmjereno na primjenu genetskih algoritama pri rješavanju ovog optimizacijskog problema, i tu je svrhu razvijen i implementiran računalni model u programskome paketu Matlab.
For the last 7 years the Republic of Croatia has been witnessing a sudden increase of distributed energy resources connecting to its power system. However, a significant problem is how to find the financial equilibrium which will satisfy the Independent Power Producers on one side and the Distribution System Operator on the other. In his scientific research the author has used capital budgeting analysis for solving the distributed generation allocation problem. The evaluation of net present values was the central tool for finding the optimal position and size of distributed generation units in the medium voltage distribution network. By simultaneously maximizing the profit of distributed generation projects and minimizing the active power losses of the medium voltage network, the interests of Independent Power Producers and the Distribution System Operator were both included in the algorithm. The power production of distributed generation and power consumption of network loads was modeled with characteristic average daily power curves with discrete hour intervals. The problem was solved using genetic algorithm, and implemented in Matlab programming environment. In the second chapter of the thesis the author has described the reasons for rapid research, development and implementation of distributed generation in many countries of the world, especially focusing on the EU. The chapter contains a brief description of different distributed generation technologies and the financial and legislative framework for their integration in the Croatian power system. The third chapter is about the transition of the once traditional and passive distribution networks into active networks containing both consumers and producers (i.e. prosumers). The underlying statement is that the voltage/thermal parameters of the network significantly change with the presence of distributed generation, making their integration an extremely complex task with many variables. The chapter stresses the importance of using the average daily power consumption/production curves, which enable the distribution network planners a more detailed and thorough perspective on distribution network operation. Capital budgeting is elaborated in the fourth chapter. Terms relevant for project management such as discounting and weighted average cost of capital are explained in detail. Also, the chapter describes the main financial characteristics of the projects and their risks. Since the capital budgeting projects are expected to generate cash flows over several years, the decision to accept or reject a project depends on the analysis of the cash flows generated by the projects and its costs. The basic formal methods in capital budgeting analysis are covered in the chapter, especially focusing on the net present value which is used as a central evaluation tool for the objective function. The fifth chapter describes the optimization problems and thier solving techniques. The exact and heuristic methods are explained along with their pros & cons. Since the genetic algorithm is used for solving the distributed generation allocation problem, its general description is the main part of this chapter. Many papers addressing the DG allocation issue are in the focus of the following chapter. Different types of objective functions have been presented. The past research based on single-criteria and multi-criteria optimization has been explained, compared and analysed. As a conclusion the compliance of voltage and thermal constraints is often not enough for distributed generation allocation, because non-optimally allocated production units may cause increased losses, voltages, reverse power flows thus effectively sterilizing the network for future connections. For that reason, the distributed generation allocation problem needs to be approached thoroughly, with all the daily/monthly/yearly network statistical measurement data taken into consideration. Renewable sources like sun and wind cannot guarantee that the nominal power output from solar parks and wind farms will be maintained in a 24 hour period. As they are stochastic primary sources, their intermittent nature must be taken into consideration while planning their integration in the network. In that way, the average daily power consumption/production curves can be formed, giving the planner a more precise and correct output results. The objective must be the simultaneous maximization of the investors' and Distribution System Operator's financial gains. The aforementioned premises are in the core of the seventh chapter. The central part of the thesis is to simultaneously maximize the net present values of distributed generation projects and loss savings in the medium voltage distribution network. The objective function consists of two parts, one involving distributed generation projects and the other involving loss savings. The net present values of distributed generation projects are calculated by discounting the annual cash flows to the present value. The time period under consideration is equal to the number of years in which the feed-in tariffs are guaranteed by the subordinate legislation. The net present value for a Distribution System Operator is calculated by discounting the annual savings to the present value. The annual savings consider the difference between annual network energy losses of the existing network and after allocation of new production units. A modified genetic algorithm was used for solving the distributed generation allocation problem. The algorithm was tested on a 20 kV distribution network consisting of 30 nodes. The results have been presented and explained at the end of the chapter. The application of the new algorithm is covered in the eight chapter. The algorithm was tested on a 20 kV distribution network without previously connected distributed generaton units. Then the algorithm was tested on the same network but with a previously connected production unit. Finally, the algorithm was tested on a network with identical input data but operated on a 10 kV level. The results have been presented, analysed and compared at the end of each subchapter. In his research the author has presented an algorithm for the allocation of distributed generation units based on financial evaluation of distributed generation projects and Distribution System Operator's savings. The allocation criterion was the maximization of the sum of their net present values based on the average daily power consumption/production curves. The algorithm presents the decision making part of project management which involves choosing the optimal size and position of future power sources prior to their connection in the network. Since the algorithm targets the overall benefits for the Independent Power Producers and the Distribution System Operator, the author hopes it will be applied for the integration of new production units in real distribution networks.