Department of Physicshttp://hdl.handle.net/11602/19232024-03-29T12:16:23Z2024-03-29T12:16:23ZComparison and evaluation of empirical and machine learning models in estimating global solar radiation in Limpopo provinceMurida, Thalukanyo Witneyhttp://hdl.handle.net/11602/25892023-11-28T09:14:40Z2023-10-05T00:00:00ZComparison and evaluation of empirical and machine learning models in estimating global solar radiation in Limpopo province
Murida, Thalukanyo Witney
This study investigated the performance of machine learning techniques as compared to the
empirical models to forecast the global solar radiation in Limpopo regions. The machine
learning techniques used in this study are Support Vector Machines, Random Forest, and
Artificial Neural Network, and the empirical models used are the Clemence and Hargreaves-
Samani models. To assess the efficiences of the machine learning models against the empirical
models, the researchers calculated and compared the models performance evaluation
using statistical equations such as Coefficient of determination, Mean Square Error, Mean
Absolute Error, and Root Mean Square Error. Calibaration was done to improve performance
of the empirical models. The present study found that machine learning techniques
perform better than the empirical models when estimating the global solar radiation in the
selected Limpopo regions.
MSc (Physics); Department of Physics
2023-10-05T00:00:00ZForecasting Minute Averaged Solar Irradiance Using Machine Learning for Solar Collector ApplicationsNemalili, Ronewa Collenhttp://hdl.handle.net/11602/24942023-06-08T14:35:19Z2023-05-19T00:00:00ZForecasting Minute Averaged Solar Irradiance Using Machine Learning for Solar Collector Applications
Nemalili, Ronewa Collen
Challenges in utilising fossil fuels for generating energy call for the use
of renewable energy. This study focuses on modelling and forecasting
solar energy and optimum tilt angle of solar energy acceptance using
historical time series data collected from one of the South African radiometric
stations, USAid Venda station in Limpopo province. In the study
we carried out a comparative analysis of Random Forest and Bayesian
linear regression in short-term forecasting of global horizontal irradiance
(GHI). To compare the predictive accuracy of the models, k-Nearest
Neighbors (KNN) and Long short-term memory (LSTM) are used as
benchmark models. The top two models with the best performances
were then used in hourly forecasting of optimum tilt angles for harvesting
solar energy. The performance measures such as MAE, MSE, and
RMSE were used and the results showed RF to have better performance
in forecasting GHI than other models, followed by the LSTM and the
third best model was the KNN whereas the BLR was the least performing
model. RF and LSTM were then used in modelling and forecasting
the tilt angles of optimal solar energy acceptance and as thus, the LSTM
outperformed the RF by a small margin.
MSc in e-Science; Department of Physics
2023-05-19T00:00:00ZFirst-principles study of Hematite (α-Fe2O3) surface structures doped with Copper (Cu), Titanium (Ti), nickel (Ni) and manganese (Mn)Mabaso, Clarence Vusihttp://hdl.handle.net/11602/24792023-06-09T06:47:50Z2023-05-19T00:00:00ZFirst-principles study of Hematite (α-Fe2O3) surface structures doped with Copper (Cu), Titanium (Ti), nickel (Ni) and manganese (Mn)
Mabaso, Clarence Vusi
Hematite has attracted research interest for many years due to its application in water splitting. Despite its attractive characters such as a reasonable optical band gap, the semiconductor is still faced with great uncertainty for the accomplishment of hematite based photoelectrochemical cells for water splitting. Doping with transition metals has shown to be a practical solution to overcome some of the limitations faced with hematite by modifying the energy band to improve its photo-electrochemical (PEC) activity. This study explored two surface structures of pure and transition metals (Ti, Cu, Ni and Mn) doped- α-Fe2O3 oriented in the directions (001) and (101). Calculations via the first principle using the density functional theory (DFT) were adopted, the results show that the doping of transition metals in α-Fe2O3 has an effect in modifying both the valence and conduction band edges. Specifically, doping Ti introduces more electrons in the conduction band and fills the unoccupied 3d states, which could improve the rate of charge transportation and likely enhance the electrical conductivity of α-Fe2O3. Doping with Mn, Ni, and Cu has effectively improved the absorption coefficient for α-Fe2O3 (001) and (101) surfaces, in the visible light region. The overall analysis of the results shows an opportunity for a successful photo-electrochemical water splitting application.
MSc (Physics); Department of Physics
2023-05-19T00:00:00ZDevelopment of a mathematical model for predicting bio-slurry temperature and subsequent gas production rate for underground brick-built biogas digester using ambient air temperature forecastNekhubvi, Vhutshilo 1st Mountaineerhttp://hdl.handle.net/11602/24052023-04-26T06:46:29Z2022-11-10T00:00:00ZDevelopment of a mathematical model for predicting bio-slurry temperature and subsequent gas production rate for underground brick-built biogas digester using ambient air temperature forecast
Nekhubvi, Vhutshilo 1st Mountaineer
Background: Heat energy is essential for the anaerobic digestion of organic materials such as household, human or agricultural waste. Many developing countries have witnessed efforts to implement anaerobic digestion technology for biogas production as a strategy to enhance energy supply and poverty eradication in rural communities. Underground, brick, and mortar built fixed dome type digesters are the most deployed small-scale biogas technology in sub-Saharan Africa (SSA) countries such as Rwanda, Ethiopia, Tanzania, Kenya, Uganda, Burkino Faso, Cameroon, Benin, Senegal, and South Africa despite their relatively high initial costs. They have a long lifespan and no moving or rusting parts involved. The basic design is compact, saves space, is well insulated, and does not need additional heating, hence suitable for developing countries. The technology is labour-intensive that involves digging the pit and constructing the structure from underground, thus creating local employment. Unlike prefabricated biogas digesters, underground, brick, and mortar-built fixed dome type digesters are more robust than the latter, with minimal gas pipes corrosion experienced.
However, little literature on this type of digesters' actual field operation and performance within the SSA context is available. The end-user must know what needs to be done and what the system's outcome is supposed to be. Besides determining parameters like total solids, volatile solids, carbon-nitrogen ratio, hydrolysis rate, organic loading rate, and hydraulic retention time, the temperature inside the digester becomes one of the metrics to evaluate the anaerobic digestion process. The digestion temperature critically affects the biogas yield, considering all other conditions unchanged. Knowing the operational temperature, one can estimate the maximum specific growth rate of the microorganisms and the biogas production rate. Prediction models for the internal operating temperature of these digesters under local conditions typical of Limpopo province of South Africa, where most of these digesters have
been installed,
are still lacking. To ordinary users in rural areas, the prediction of the possible 'duration of use,' for example, the duration of continuous cooking, is essential. However, regardless of fulfilling all other operational requirements to predict daily gas production, internal digester temperature remains the missing link to having a complete set for a quick and easy gas yield estimation.
Aim of the study: The overall objective was to develop a locally applicable model for predicting the bio-slurry operating temperature of underground brick-built domestic size biogas digesters. The work established a correlation of ambient air temperature with the slurry temperature inside the digester using a heat transfer mechanism through the media between the fermenting slurry and the ambient air.
Methodology: A thermodynamic study of a small-scale fixed-dome Deenbandhu biogas digester model was performed by monitoring the digester's temperature and surroundings. The K-type chromium-nickel temperature sensors with a sensitivity of 41 μV/°C and a response time of 0.8 s in liquids were positioned at the centre of the digester to measure the slurry temperature. Another temperature sensor was placed 2.0 m above the ground to measure ambient air temperature. The sensors were connected to the data logger and programmed to record temperature readings every second, automatically averaged hourly and daily. The soil surface heat flux was computed using Fourier's law of heat conduction to strengthen the model.
Results: The average daily bio-slurry temperature of the digesters ranged between psychrophilic and mesophilic ranges. The results show a strong correlation between bio-slurry and ambient air temperature. A strong correlation was obtained between the measured and predicted temperature of the fermenting slurry inside the digester with a ()Pr|t|>value less than 2e-16 ***, showing that the model is most significant. A Q-Q plot was also used to measure the importance of each observation to the regression.
Conclusion: The developed models can accurately estimate the bio-slurry temperature inside the digester using local ambient air temperature data. The set equation adds value as input to the research of small-scale household biogas digesters. Furthermore, the biogas production rate was calculated using data on predicted slurry temperature. It was found that the biogas production rate is satisfactory, given the condition of the study area. The biogas production rate varies from as low as 0.18 m3m-3d-1 during the cold month to 0.48 m3m-3d-1 during the warmest month. Temperatures above 20 ℃ were more conducive to a high biogas production rate.
PhD (Physics)
2022-11-10T00:00:00Z