A carregar...
Publicação
Avaliação de situações de separação de mercado no MIBEL e determinação de causas possíveis
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 5.76 MB | Adobe PDF |
Orientador(es)
Resumo(s)
O Mercado Ibérico de Energia Elétrica, MIBEL, permite a integração dos sistemas
elétricos português e espanhol num mercado de eletricidade comum. Devido às limitações
físicas das interligações entre as zonas de mercado destes dois países, nem sempre é
possível realizar a totalidade das trocas determinadas em mercado, pois a capacidade de
interligação é superada. Nestas situações, é aplicado o mecanismo de gestão conjunta das
interligações entre Portugal e Espanha, conhecido como separação de mercado, que resulta
no cálculo dos preços de eletricidade com base nas curvas de procura e oferta especificas
a cada zona de mercado, resultando em preços diferentes. O principal objetivo desta
dissertação é o de identificar as principais causas de separação de mercado considerando
o contexto atual do MIBEL.
É feita uma análise inicial da evolução de separação de mercado e são identificados
alguns padrões de ocorrência para o período de 2017 a 2023. Foram também estudados
os graus de correlação entre as variáveis de estudo e a diferença de preços entre Portugal
e Espanha em momentos de separação de mercado.
A separação de mercado é modelada como uma variável binária alvo, que conjuntamente
com os dados de produção de energia elétrica discriminados por tecnologia,
consumos nacionais e capacidades de interligação, são utilizados na construção de três
modelos explicativos distintos: Regressão Logística, Random Forest e Rede Neuronal
Artificial. Utilizam-se técnicas de Explainable Machine Learning, sempre que adequado,
como forma de extrair conhecimento acerca dos determinantes da separação de mercado
destes modelos.
Por fim são analisados os resultados obtidos e são feitas algumas recomendações para
a redução da separação de mercado tendo em conta os futuros desenvolvimentos dos
sistemas elétricos português e espanhol.
The Iberian Electricity Market, MIBEL, allows the Portuguese and Spanish electricity systems to be integrated into a common electricity market. Due to the physical limitations of the interconnections between the market areas of these two countries, it is not always possible to realise all the exchanges determined in the market, as the interconnection capacity is exceeded. In these situations, the mechanism for joint management of the interconnections between Portugal and Spain, known as market splitting, is applied, which results in electricity prices being calculated on the basis of demand and supply curves specific to each market area, resulting in different prices. The main objective of this dissertation is to identify the main causes of market splitting in the current context of MIBEL. An initial analysis of the evolution of market splitting is made and some patterns of occurrence are identified for the period of 2017 to 2023. The degrees of correlation between the study variables and the price difference between Portugal and Spain at times of market splitting were also studied. Market splitting is modelled as a binary target variable, which together with electricity generation data broken down by technology, national demand and interconnection capacities, is used to build three different explanatory models: Logistic Regression, Random Forest and Artificial Neural Network. Where appropriate, Explainable Machine Learning techniques are used to extract knowledge about the determinants of market splitting from these models. Finally, the results obtained are analysed and some recommendations are made for reducing market splitting, taking into account future developments in the Portuguese and Spanish electricity systems.
The Iberian Electricity Market, MIBEL, allows the Portuguese and Spanish electricity systems to be integrated into a common electricity market. Due to the physical limitations of the interconnections between the market areas of these two countries, it is not always possible to realise all the exchanges determined in the market, as the interconnection capacity is exceeded. In these situations, the mechanism for joint management of the interconnections between Portugal and Spain, known as market splitting, is applied, which results in electricity prices being calculated on the basis of demand and supply curves specific to each market area, resulting in different prices. The main objective of this dissertation is to identify the main causes of market splitting in the current context of MIBEL. An initial analysis of the evolution of market splitting is made and some patterns of occurrence are identified for the period of 2017 to 2023. The degrees of correlation between the study variables and the price difference between Portugal and Spain at times of market splitting were also studied. Market splitting is modelled as a binary target variable, which together with electricity generation data broken down by technology, national demand and interconnection capacities, is used to build three different explanatory models: Logistic Regression, Random Forest and Artificial Neural Network. Where appropriate, Explainable Machine Learning techniques are used to extract knowledge about the determinants of market splitting from these models. Finally, the results obtained are analysed and some recommendations are made for reducing market splitting, taking into account future developments in the Portuguese and Spanish electricity systems.