Bayesian belief network modelling of the challenges associated with hybrid solar-diesel electricity from the end users’ perspective in Bugala Island in Uganda

Abstract

Abstract Background A hybrid solar-diesel energy system (HSDES) was installed on Bugala Island located on Lake Victioria in Uganda to increase access to clean, affordable, reliable, and sustainable energy, which is essential for protecting the environment while improving living conditions, human health, and local economic development. However, the challenges associated with HSDES generated electricity from the end-users’ perspective are still unknown. Since these challenges are complex, stochastic, nonlinear, and multidimensional, this study aimed at utilizing a Bayesian belief network (BBN) modelling approach to identify, and rank the challenges experienced by energy consumers on the Island in an intergrated format. Methods A cross-sectional research design was employed where a total of 237 randomly selected respondents were involved in this study. Quantitative data were gathered using semi-structured questionnaires. The BBN model was developed basing on well established guidelines and modelling protocols. Using the K-fold partitioning approach (K = 2), the casefile (n = 237) was randomly partitioned into a training portion (70%, n = 165) used to populate the model, and test dataset (30%, n = 72), with which accuracy was assessed. Model accuracy was evaluated using metrics of sensitivity, and predication performance. Results The developed model was 81.7% accurate in predicting the challenges correctly. The model's spherical payoff was 0.87 with the logarithmic and quadratic losses of 0.33 and 0.23, respectively indicating a strong predictive power and the model’s classification power. The probability of households to experience significant challenges associated with HSDES generated electricity was 30.6%. The rate of damage of electric appliances increased by 15.6 percentage points. Cases of faulty electric meters increased by 25.8 percentage points. The unit cost of electricity, and the cost of power connection also increased by 1.9 and5.5 percentage points respectively. Cases of electrocution increased by 2.4 percentage points while load shedding increased by 4.7 percentage points. Conclusion This study provides the key top ranked challenges that should be given priority in an effort to expand hybrid energy on Bugala Island. We conclude that BBN modelling approach is a promising tool in the field of hybrid renewable energy systems on Islands with potential applications due to its versatility.