Modelling electricity load scheduling and retailer decision-making

Abstract

Electricity is critical to the economic and social development of humanity. Significant effort has been spent on the effective management thereof and with the growth of the renewable energy sector, traditionally regulated markets are no longer sufficient. This has resulted in privatization of the sector over the last three decades, and has largely been met with success internationally. South Africa however, continues to suffer rolling black-outs and rising energy costs. Many attribute this to the closed system under which the country operates. In order for there to be sufficient buy-in for policy-change, key stakeholders such as the consumer and retailer must be made aware of the new reality under which they would operate, the factors that would affect their interests, and the extent to which they would be affected. Furthermore, the conflicting objectives of these parties must also be addressed through their simultaneous achievement, taken to be social welfare. This dissertation satisfied these aims by creating an accurate depiction of the locally unique consumer and retailer?s realities through the development of operations research models. The resident?s problem was modelled as a load scheduling one which considered the vastly divergent socio-economic status of South Africans and how this affects their energy consumption patterns. The spot market dynamics that a retailer is confronted with was modelled as a three-regime Markov switching model. Because social welfare was the overwhelming interest of the study, a novel problem formulation was proposed to combine the resident?s interests in reducing bill payments and inconvenience levels, and the utility?s interest in increasing revenues. The developed models and problem formulation were applied to South African scheduling data for residents operating under a fixed rate tariff. It was found that, under the guidelines of Eskom?s pricing boundaries, the relationship of the consumer?s price elasticity relative to the retailer was not a linear one. Social welfare was found to be a function of this relationship, and static tariffs that achieved optimal social welfare at varying degrees of relative price elasticity were identified. It was noted that insufficient research has been conducted on validating the effect of the retail tariff on the resident and utility. Furthermore, this effect varies from one society to the next and is dependent on factors such as consumer attitudes and electricity profit margins. Time-varying tariffs increased model complexity but are capable of achieving demand response which are believed to broaden the interests of the retailer and consumer. A trail-and-error algorithm was proposed as an appropriate tool for demonstrating the effects of demand responsiveness under a time-of-use (TOU) tariff. This was applied to the South African context with the inclusion of the novel problem formulation. The novelty of this thesis is four-fold: firstly, a problem formulation that captures social welfare, which has previously not been considered in literature, is proposed. Secondly, the assumption of most works in this field that the effect of retail tariff changes on the consumer and retailer are the same, is disproved. In fact, this relative sensitivity is shown to be far greater for the resident than for the utility. Thirdly, a three-regime Markov switching model is successfully applied to the Australian market with no restrictions on the transition probability matrix. Finally, initial computations for this unique perspective on the problem are conducted with a trial-and-error algorithm and findings will certainly assist in guiding future research.