Browsing by Author "Firehiwot Girma"

Now showing 1 - 1 of 1
  • No Thumbnail Available
    Item
    Hydropower Generation and Operation Planning For Ethiopia
    (Addis Ababa University, 2021-12) Firehiwot Girma; Getachew Bekele (PhD); Mikael Ameline (Prof.)
    The Ethiopian power system is highly dominated by hydropower plants. Almost 90% of the generation is covered by hydropower. Although the total generation capacity in the power system is sufficient enough to cover the peak demand, it is common to see load shedding and power rationing in the country, especially in the dry seasons of the year. The absence of a suitable and appropriate generation planning tool makes the current planning dependent on historical generation patterns. Recent literature has shown different models for hydropower planning for long-term and short-term operation in a deterministic, stochastic, stochastic dynamic, etc., with different levels of details and mathematical formulations. However, most of the models are concerned with profit maximization and cost minimization in a competitive electricity market. In addition, no previous studies have been conducted, and no models have been developed for the planning of the unique Ethiopian power system. The Ethiopian power system is a system dominated by hydropower generation, which is dependent on seasonal rainfall. The electricity market is a vertically integrated market where the government determines the electricity price. Therefore, there is no price uncertainty and less concern about cost or profit. The primary concern for the planning of the Ethiopian power system will be the proper scheduling of the power plants to use the stored water in the rainy season through the dry season with minimum load shedding to increase system reliability while keeping the balance between load shedding now and in the future. In principle, load shedding can be avoided by using all the water available right now, but then, if there is a poor rainy season with low inflow the following year, the power system may get into a massive problem supplying the demand for that year. Before the market deregulation, the electricity market in developed European countries was almost similar to what is practiced in Ethiopia now. The difference is they have much more details of measurements and statistics about inflows, run time between reservoirs, etc. This thesis develops different hydropower planning tools, including deterministic and stochastic, risk-neutral, and risk-averse models for the Ethiopian power system based on the limited available data, intending to utilize the water stored in the rainy season throughout the year with minimum load shedding. It further studies and tests the models in a rolling horizon framework for long-term operation.The Methodology used to develop the hydropower planning tool is, first, all the necessary data for the planning is collected from Ethiopian Electric Power (EEP), inflow is scaled from the mean annual energy (MAE) of each reservoir using the publicly available precipitation data from NASA. Then, the deterministic model is developed and compared with the historical generation data. In hydropower generation, inflow is an uncertain stochastic process. To consider the uncertainties in the inflow, the deterministic model is further developed into a two-stage stochastic model. To run the stochastic model, we formulate a method to prepare a synthetic historical inflow series from the available data on hand and derive a method to estimate the stochastic process that mimics the synthetic historical series. We use Monte Carlo simulation to generate random inflow scenarios from the estimated stochastic process. The stochastic planning model is then tested both in a risk-neutral and a risk-averse version. We use Conditional Value-at-Risk (CVaR) risk measure to develop the risk-averse model. Finally, the performance of the models developed is compared using a rolling horizon framework for a one-year planning period. The results show that the Ethiopian power system has a great deal of flexibility to be operated more efficiently to minimize load shedding. The results also show that by using stochastic models, we can better manage the water in the reservoirs in the form of slightly lower load shedding without compromising the energy we reserve for the next planning period. We could also avoid large load shedding events so that the load shedding is evenly distributed throughout the year instead of having massive load shedding in a short period, which could be very valuable when we have higher load demand in Ethiopia. When testing the hydropower planning tools for current load demand, there is a very good generation capacity to supply the demanded load. However, there was significant load shedding in the actual operation, even though the planning model suggested no need for load shedding. It is concluded that it will be an improvement if the planning is supported by stochastic planning tools instead of using the method depending on historical data.