Probabilistic distributions for modelling seasonal load profiles of commercial areas in south africa

One of the most significant commodities of today's world is energy. Energy usage depends on various factors such as season, day of the week, temperature etc. It is imperative that the distribution, transmission, and generation of electricity is effective while equally producing required results to electricity customers. With an expectation for increasing power outages in South Africa in the nearest future, there is a renewed focused on electricity distribution and consumption. This paper examines the electric load profile at a commercial location in Johannesburg, South Africa, for which the overall dataset (in KWh) is classified into four seasonal regimes: summer, spring, winter, and autumn. Two probabilistic models - normal and lognormal distributions - are applied to investigate the medium-term behaviour of the time series dataset over a period of two years, between 2019 and 2020. Results from this investigation suggest that normal distribution gives a better approximation to the seasonal datasets, except during the spring season. The lognormal distribution is observed to give minimal fitting errors during the spring season. Additionally, the load profile during summer and spring seasons are observed to exhibit similar characteristics, likewise, both autumn and winter seasons are found to exhibit the same trend for the same period.