George Irungu

Dr. George Irungu

Tshwane University of Technology in Pretoria
orcid

Research Focus & Metrics

Academic Career Length

8Years (2019-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
28 May 2022 in Journal of Electrical and Computer Engineering
Verified
Journal Article
Open access
PDF file
HOHampfrey Odero
Hampfrey Odero
CWCyrus Wekesa
Cyrus Wekesa
GIGeorge Irungu
George Irungu
Kenya is experiencing a fast increase in grid-connected intermittent renewable energy sources (RESs) to meet its increased power demand, and at the same time be able to fulfill its Paris Agreement obligations of abating greenhouse gas emissions. For instance, Kenya has 102 MW of grid-tied solar power and 410 MW of grid-tied wind power. However, these sources are very intermittent with low predictability. Thus, after their installation and integration into the grid, they impose a new challenge for the secure, reliable, and economic operation of the system. To mitigate these and to ensure proper planning of the system operations, accurate and faster prediction of the generation output of the wind energy resources and optimal design and sizing of storage for the large-scale wind energy integration into the grid are of paramount importance. Artificial intelligence (AI) and metaheuristic techniques have proven to be efficient and robust in offering solutions to complex nonlinear prediction and optimization problems. Therefore, this study aims to utilize backpropagation neural network (BPNN) algorithm to conduct hourly prediction of the generation output of Lake Turkana Wind Power Plant (LTWPP), a 310 MW plant connected to the Kenyan power grid, and optimally size its battery energy storage system (BESS) using genetic algorithm (GA) to guarantee its dispatchability. The historical weather data, namely wind speed, ambient temperature, relative humidity, wind direction, and generation output from LTWPP, are employed in the training, testing, and validation of the neural network. LTWPP and BESS are modelled in MATLAB R2016a software. Thereafter, the developed BPNN and GA algorithms are applied to the modelled systems to predict the wind output and optimize the storage system, respectively. BESS optimization with neural prediction reduces the BESS capacity and investment costs by 59.82%, while the overall dispatchability of LTWPP is increased from 73.36% to 90.14%, hence enabling the farm to meet its allowable loss of power supply probability (LPSP) index of 0.1 while guaranteeing its dispatchability.

9 October 2019 in Engineering, Technology & Applied Science Research
Verified
Journal Article
Open access
PDF file
SOS. H. Eisa Osman
S. H. Eisa Osman
GIGeorge Irungu
George Irungu
DMD. K. Murage
D. K. Murage
Induction power generators are the most popular wind energy conversion systems (WECS) because they do not require synchronization units. However, they usually draw a huge quantity of reactive power during disturbances. Hence, incorporating wind power into power networks may cause voltage instability. This paper presents the usage of STATCOM and SSSC FACTS devices for voltage stability enhancement of a distribution network with a squirrel cage induction generator (SCIG) wind power turbine. The continuation power flow (CPF) approach is utilized as a tool to determine the most suitable position of SCIG in the system. Also, voltage stability indices (FVSI and Lmn) are employed to estimate the stability margin of the system by figuring the weakest transmission lines and buses in order to locate the appropriate position where the FACTS devices should be installed. A comparison of the suitability of the FACTS devices to restore system stability was evaluated under 3-phase fault conditions. The results illustrated that STATCOM behaves better than SSSC when the system is restoring from a fault. Simulations and voltage stability assessment were carried out on the IEEE 14 bus test scheme using the PSAT simulation software package.

2 February 2025 in Engineering, Technology & Applied Science Research
Verified
Journal Article
Open access
PDF file
PDParach Daniel Deng
Parach Daniel Deng
GIGeorge Irungu
George Irungu
Josiah Lange MundaJosiah Lange Munda
Josiah Lange Munda
Electrical distribution systems are undergoing significant modifications since the application of new technologies. New possibilities for automated, dependable, and efficient electrical power grids have been made possible by the technological advancement. While new technologies might improve electrical network performance and offer creative solutions to future network difficulties, they can also have unintended consequences that need to be carefully studied and considered. A recent technological advancement that enhances power grid performance is Distributed Generation (DG). While DG unit integration has measurable benefits for electrical grids, its significant effects on power network protection systems create many questions and difficulties about the proper way to identify and isolate distribution network faults. The DLANN-based approach looks into the ways the integration of DGs affects fault identification and location. This method involves two steps: first, three-phase currents are constantly analyzed for detection, and Discrete Wavelet Transform (DWT) is utilized to extract the currents' features. The second step is classification employing Artificial Neural Networks (ANNs) to pinpoint the defective stages. Counting the shorted phases will reveal the sort of short circuit. The MATLAB programming environment is utilized in the development of the fault identification and classification technique. The fault type (one, two, or three phases), fault resistance, fault location bus, fault distance, and the DG type (upstream or downstream) are all considered. The methodology is used on a modified IEEE 34-bus test system, and four scenarios, one with combined DGs units, one with IBDGs, one with SBDGs, and one without DGs, are modeled. As per the simulation results, 100% fault detection and classification accuracy were obtained, whereas the average fault location accuracy attained without DGs, with IBDGs, SBDGs and combined DGs for selected nodes were 99.94%, 99.91%, 99.86%, and 99.88%, respectively

6 January 2026 in Journal of Electrical and Computer Engineering
Verified
Journal Article
Open access
JNJulius Ndirangu
Julius Ndirangu
JNJohn Nderu
John Nderu
GIGeorge Irungu
George Irungu
The increasing integration of renewable energy systems (RESs) such as solar and wind into modern power grids using converter‐based, inertia‐less interfaces poses serious frequency stability challenges. This is due to reduced system inertia conventionally provided by synchronous generators. This paper proposes an innovative solution using an adaptive grid‐forming controller (AGFC), designed to emulate and adapt virtual inertia and damping in real time. The AGFC uniquely combines grid‐forming virtual inertia damping control with an adaptive DC link voltage regulation loop, coordinated by a genetic algorithm–optimized proportional integral (GA‐PI) controller. Key innovation lies in employing the GA‐PI control to dynamically manage DC link energy exchange during grid disturbances. Through optimized GA‐PI tuning, rapid adjustment of control parameters is achieved to maintain system stability during sudden load changes. MATLAB/Simulink simulations demonstrate that the proposed AGFC substantially mitigates frequency deviation, rate of change of frequency, and frequency settling time under transient conditions. Compared with conventional grid‐forming control without adaptive DC link management, the designed AGFC depicts superior resilience and adaptability to load disturbances. It enables inverters to autonomously establish voltage and frequency references while providing synchronous generator‐like stabilization. The results demonstrate that the proposed framework enhances the grid’s inertial response and dynamic stability for converter‐interfaced RES. Future research will explore scaling the AGFC for multiarea grids and applying advanced parameter optimization to larger systems. This is expected to further improve system‐wide frequency regulation and stability.

22 April 2026 in IEEE Access
Verified
Journal Article
Open access
PDParach Daniel Deng
Parach Daniel Deng
GIGeorge Irungu
George Irungu
Josiah Lange MundaJosiah Lange Munda
Josiah Lange Munda
This research describes a deep learning-based artificial neural network (ANN) approach for the detection, classification, and localization of series faults (open conductor faults) in power distribution networks. The ANN is trained using the basic elements of the voltage and current signals that are recovered at the relaying point as input attributes. After an ANN has been trained, it must be tested for fault scenarios that were not encountered during training. The proposed DLANN-based fault detector is evaluated in MATLAB/Simulink for all types of series faults in distribution lines with variations in fault location and fault inception time of 0.2 seconds, including one conductor open fault, two open conductor faults, and three open conductor faults. A modified IEEE 34-bus test system is used to test the methodology, and two scenarios one with and one without distributed generation units are modelled in the MATLAB environment. The detection and classification accuracy for the series faults were both achieved at 100% and the fault location accuracy for series fault with and without DGs were obtained as 99.7% and 99.5% respectively.

15 May 2026 in Preprints.org
Verified
Preprint
Open access
PDF file
CMChristian Kakule Mathe
Christian Kakule Mathe
GIGeorge Irungu
George Irungu
AIAlice Ikuzwe
Alice Ikuzwe
The increasing penetration of renewable energy sources demands dispatch strategies that balance technical reliability, environmental sustainability, and economic efficiency. While Hybrid Photovoltaic/Hydro/Diesel/Battery Energy Storage (BESS) systems have been studied, most existing works focus on single-objective optimization or genetic multi-objective trade-offs without explicit integration of global sustainability thresholds. This study introduces a novel policy-embedded Mixed Integer Nonlinear Programming (MINLP) dispatch framework that embeds policy-aligned constraints (losses ≤ 8%, CO₂ reduction ≥ 40%, and cost reduction ≥ 20%) directly into the optimization model of the IEEE 30-bus system. Unlike prior studies, the framework establishes a replicable benchmark for hybrid generator placement and sizing, combining renewable-first dispatch logic with explicit emission and cost caps.Results demonstrate that policy thresholds are achievable within technical feasibility, with losses halved, emissions reduced by over 40%, and costs lowered by 20%. Pareto frontier analysis reveals that global policy targets coincide with the frontier of achievable trade-offs, providing new evidence that sustainability agendas can be operationalized in dispatch optimization. This contribution advances hybrid system research by bridging technical modeling with global energy policy, offering actionable insights for grid operators, policymakers, and researchers. By systematically locating PV+BESS at Bus 19/30, Hydropower at Bus 6/11 and Diesel at Bus 2/5, the study provides a reproducible design logic that future researchers can adopt. This benchmark moves beyond abstract optimization to offer a practical system design contribution.