Institute of Technology
Permanent URI for this communityhttps://etd.hu.edu.et/handle/123456789/66
The Institute of Technology focuses on education, research, and innovation
in engineering, technology, and applied sciences to support sustainable development.
Browse
3 results
Search Results
Item DISTRIBUTION NETWORK OPTIMIZATION BY OPTIMAL SIZING AND PLACEMENT OF D-STATCOM USING TEACHING AND LEARNING BASED OPTIMIZATION ALGORITHM (CASE STUDY: YIRGALEM SUBSTATION)(Hawassa University, 2021-10-18) AZMERAW ARGAWDistribution system is part of an electric power system which links the high voltage transmission networks with the end consumers. This work offers the way of improving the performance of the distribution network by improving voltage profile and reduction of power loss via injecting reactive power through the network. Optimal siting and sizing of custom power devices in power distribution networks maximizes voltage profile, compensates reactive power, minimizes power loss and enhances voltage profile. The search for optimal size and locations of these devices in radial distribution networks is challenging and requiring robust scheduling. This study is conducted with a focus on Aposto feeder of Yirgalem distribution network. The voltage profiles of most buses are not in an acceptable range, and the voltage stability index of the buses shows that network is prone to voltage stability problem. In this study, it is aimed to find the best optimal D-STATCOM sizing and placement by using Teaching and Learning Based Optimization (TLBO). Results obtained have been compared with those of the conventional optimization techniques reported in literature. For the Aposto feeder 62-bus network, the optimal location and size of D-STATCOM were determined at bus 38 with 1019.18 𝑘𝑉𝑎𝑟, at bus 28 with 942.96 𝑘𝑉𝑎𝑟, at bus 39 with 1074 𝑘𝑉𝑎𝑟 and at bus 25 with 1184 𝑘𝑉𝑎𝑟 by the GA, PSO, GREY WOLF and WHALE OPTIMIZATION method respectively. While the TLBO approach obtained the optimal site and size of the D-STATCOM in the network to be bus 51 and 871.4 𝐾𝑉𝑎𝑟 at normal load condition. As stated, the TLBO method performs better in terms of reducing both real and reactive power losses. The real power loss percentage reduction of the test system is 27.09%, 69.19% and 70.24% whereas the reactive power loss percentage reduction is 30.97%, 68.85% and 69.85% for light load, normal load, and heavy load respectively. Also, the minimum voltage level in the worst case is significantly enhanced from 0.93pu to 0.988pu. The model has been formulated to minimize the total cost of the network by determining the optima of the substation locations and power, the load transfers between the demand centers, the feeder routes and the load flow in the network subject to a set of constraints. As per the economic evaluations, the proposed solution is cost-effective. In this research D-STATCOM control is developed based on artificial intelligent (AI) using artificial neural network (ANN), which depends on optimum values obtained by TLBO. Generally, the simulation results show that the proposed technique is effective to maintain all buses voltage magnitude within the IEEE acceptable limit and to reduce power losses significantlyItem “DETECTION, CLASSIFICATION AND MITIGATION OF POWER QUALITY DISTURBANCE: A CASE STUDY ON THE 15KV DISTRIBUTION FEEDER 6(R4-G5) AT HAWASSA SUBSTATION”(Hawassa University, 2025-12-19) DAWIT DABAModern power systems face several difficulties due to power quality disturbances, such as voltage sags and swells, which call for effective detection, classification, and mitigation techniques. In order to fully address these problems, this study offers an integrated strategy that combines modern machine learning methods with optimization techniques. Artificial Neural Networks (ANN) trained and refined with MATLAB's Classification Learner Toolbox are used for detection and classification after features are extracted from voltage/current signals. This thesis was carried out on one of the Hawassa Feeder-6 (R4-G5) 15 kV distribution feeders utilizing distribution network analysis and MATLAB simulation. The purpose of power flow analysis is to ascertain the active and reactive power flows on the distribution lines, as well as the voltage magnitude and phase angle at each bus (node) in the system. Power quality disturbances are divided into four distinct wavelet filter levels using Debechesh-4 (Db-4). This enables the improvement of an approximate and detailed coefficient distribution in addition to the extraction of features such as the mean, maximum, and lowest values of the disturbances for power quality disruptions. The classification efficacy of neural networks (ANN) and support vector machines (SVMs) is 100%. Finally, a dynamic voltage restorer (DVR) is positioned optimally using the Grasshopper Optimization (GOA) techniques. Power loss decreased from 1913.3 kW (active) and 1202.4 kVAR (reactive) to 295.534 kW and 261.803 kVAR when GOA was used, and the voltage profile was increased from 70% to 98.5% and lowered the voltage swell from 110% to 98%, and also, by applying different kinds of faults, easily tested the voltage sag and swell by using DVR integrated with wavelet transforms algorithm. This reduces the possible influence on delicate systems and equipment while simultaneously enhancing the power supply's qualityItem ENGINEERING CHARACTERIZATION OF EXPANSIVE SOIL IN ARBAMINCH TOWN (CASE STUDY OF ARBA MINCH TOWN(Hawassa University, 2020-06-21) KALKIDAN ALEMAYEHU GEBREMARIAMExpansive soils, which are considered as problematic soils, have worldwide distribution and hence, their proper identification and characterization becomes an absolute requirement in the perspective of the present day geotechnical engineering practice. Expansive soils experience significant volume change associated with changes in moisture contents, these volume changes can either be in the form of swell or in the form of shrinkage; and this is why they are sometime known as Swell/Shrink soils. Arba Minch town experiences more damage from expansive soils throughout the year. It is known that the magnitude of swelling for expansive soil varies with environmental conditions. This study intends to characterize expansive soil found in Arba Minch town and develop correlation between index properties and swelling pressure (Ps) for the study area. The soil specimen were obtained from ten different test pits from which index properties, engineering properties and swell-consolidation tests were conducted according to American Society for Testing Materials (ASTM) standard. Tested soil samples from the study area have been found to meet the diagnostic criteria for expansive soils, having LL in the range from 96% to 120%, PL ranges from 36% to 51%, free swell ranges from 97.5% to 155%, PI from 49% to 77%, GS from 2.61 to 2.83, Clay content ranges between 35.18% and 48.94%. The laboratory results revealed that the study area has plastic-behavior. In this study, efforts were made to develop Artificial Neural Network (ANN) and Multiple Regression Analysis (MRA) models that can be employed for estimating swelling pressure. Equation with high regression coefficient has been selected to predict the swelling pressure. Comparison between two software (ANN and SPSS) was conducted, ANN resulted good prediction than SPSS. Further, different parameters were used to develop the prediction model and among those Atterberg’s Limit, Dry density and Moisture content result good coefficient of correlation. Furthermore, Index properties which are used for establishing the model can be conducted easily in soil laboratory without any tedious and time consuming procedure