O presente estudo considera a aplicação do modelo SISAGUA de simulação matemática e de otimização para a operação de sistemas de reservatórios integrados em sistemas complexos para o abastecimento de água. O SISAGUA utiliza a programação não linear inteira mista (PNLIM) com os objetivos de evitar ou minimizar racionamentos, equilibrar a distribuição dos armazenamentos em sistemas com múltiplos reservatórios e minimizar os custos de operação. A metodologia de otimização foi aplicada para o sistema produtor de água da Região Metropolitana de São Paulo (RMSP), que enfrenta a crise hídrica diante de um cenário de estiagem em 2013-2015, o pior na série histórica dos últimos 85 anos. Trata-se de uma região com 20,4 milhões de habitantes. O sistema é formado por oito sistemas produtores parcialmente integrados e operados pela Sabesp (Companhia de Saneamento do Estado de São Paulo). A RMSP é uma região com alta densidade demográfica, localizada na Bacia Hidrográfica do Alto Tietê e caracterizada pela baixa disponibilidade hídrica per capita. Foi abordada a possibilidade de considerar a evaporação durante as simulações, e a aplicação de uma regra de racionamento contínua nos reservatórios, que transforma a formulação do problema em programação não linear (PNL). A evaporação se mostrou pouco representativa em relação a vazão de atendimento à demanda, com cerca de 1% da vazão. Se por um lado uma vazão desta magnitude pode contribuir em um cenário crítico, por outro essa ordem de grandeza pode ser comparada às incertezas de medições ou previsões de afluências. O teste de sensibilidade das diferentes taxas de racionamento em função do volume armazenado permite analisar o tempo de resposta de cada sistema. A variação do tempo de recuperação, porém, não se mostrou muito significativo. / The current study considers the mathematical simulation and optimization model SISAGUA applied to operation of complex multireservoir systems for water supply. The SISAGUA model uses mixed integer nonlinear programming (MINLP) with objectives of avoid or minimize shortages, balance storage distribution in multireservoir systems and minimize operation costs. The optimization methodology was applied in the water supply system from São Paulo Metropolitan Region, which faces a water crisis in a drought scenario in 2013-2015, the worst in the last 85 years historical series. It is a region with 20.4 million inhabitants, and the system consists of eight partially integrated supply systems operated by Sabesp (Sanitation Company of Sao Paulo State). The metropolitan region presents a high population density, located in the Upper Tiete hydrographic basin, characterized by low water availability per capita. It was discussed the possibility of considering evaporation during simulations, and the application of a continuous hedging rule in the reservoirs which modifies the mathematical formulation to nonlinear programming (NLP). Evaporation proved barely representative in relation to demand flow, with about 1% of the flow. On one hand, a flow rate of this magnitude may be considered in a critical scenario, on the other hand, this order of magnitude can be compared to the uncertainties of measurement or inflow forecasts. The sensitivity test of different rationing rates depending on the stored volume can analyze the time of response of each system. The change in recovery time, however, was not very significant.

Hydropower is the most important domestic energy source of Turkey. Thus, wise planning and development of the unused hydropower potential of the country is vital. There are many hydroelectric power plants under planning stage in our country. Altiparmak HEPP is one of them. General Directorate of Electrical Power Resources Survey and Development Administration (EIE) and ANC Enerji conducted two separate feasibility studies for Altiparmak HEPP in 2001 and 2009, respectively. Traditionally, the energy income calculations for HEPPs are based on DSI or EIE Methods in Turkey. Both of these methods evaluate the firm and the secondary energy generations separately. Besides they use fixed prices for these two types of energies. However, hourly electricity prices are used for electricity trading in Turkey. A detailed economic analysis of Altiparmak HEPP is conducted in this study. The economic analysis included various factors, such as tailwater level change, varying operating levels for different seasons and precipitation and evaporation amounts which are not conventionally included in feasibility studies. Moreover, the energy income calculations are conducted with four different methods, the DSI Method, the EIE Method, the ANC Method and the Variable Price Method (VPM). The VPM is developed in this study and it allows utilization of hourly electricity prices in calculating energy income of the HEPP. To shed some light on how hourly electricity prices develop, this thesis includes a chapter on the electricity market which explains the details of electricity trading in our country after the Electricity Market Balancing and Settlement Regulation became active in 2009.

O presente estudo considera a aplicação do modelo SISAGUA de simulação matemática e de otimização para a operação de sistemas de reservatórios integrados em sistemas complexos para o abastecimento de água. O SISAGUA utiliza a programação não linear inteira mista (PNLIM) com os objetivos de evitar ou minimizar racionamentos, equilibrar a distribuição dos armazenamentos em sistemas com múltiplos reservatórios e minimizar os custos de operação. A metodologia de otimização foi aplicada para o sistema produtor de água da Região Metropolitana de São Paulo (RMSP), que enfrenta a crise hídrica diante de um cenário de estiagem em 2013-2015, o pior na série histórica dos últimos 85 anos. Trata-se de uma região com 20,4 milhões de habitantes. O sistema é formado por oito sistemas produtores parcialmente integrados e operados pela Sabesp (Companhia de Saneamento do Estado de São Paulo). A RMSP é uma região com alta densidade demográfica, localizada na Bacia Hidrográfica do Alto Tietê e caracterizada pela baixa disponibilidade hídrica per capita. Foi abordada a possibilidade de considerar a evaporação durante as simulações, e a aplicação de uma regra de racionamento contínua nos reservatórios, que transforma a formulação do problema em programação não linear (PNL). A evaporação se mostrou pouco representativa em relação a vazão de atendimento à demanda, com cerca de 1% da vazão. Se por um lado uma vazão desta magnitude pode contribuir em um cenário crítico, por outro essa ordem de grandeza pode ser comparada às incertezas de medições ou previsões de afluências. O teste de sensibilidade das diferentes taxas de racionamento em função do volume armazenado permite analisar o tempo de resposta de cada sistema. A variação do tempo de recuperação, porém, não se mostrou muito significativo. / The current study considers the mathematical simulation and optimization model SISAGUA applied to operation of complex multireservoir systems for water supply. The SISAGUA model uses mixed integer nonlinear programming (MINLP) with objectives of avoid or minimize shortages, balance storage distribution in multireservoir systems and minimize operation costs. The optimization methodology was applied in the water supply system from São Paulo Metropolitan Region, which faces a water crisis in a drought scenario in 2013-2015, the worst in the last 85 years historical series. It is a region with 20.4 million inhabitants, and the system consists of eight partially integrated supply systems operated by Sabesp (Sanitation Company of Sao Paulo State). The metropolitan region presents a high population density, located in the Upper Tiete hydrographic basin, characterized by low water availability per capita. It was discussed the possibility of considering evaporation during simulations, and the application of a continuous hedging rule in the reservoirs which modifies the mathematical formulation to nonlinear programming (NLP). Evaporation proved barely representative in relation to demand flow, with about 1% of the flow. On one hand, a flow rate of this magnitude may be considered in a critical scenario, on the other hand, this order of magnitude can be compared to the uncertainties of measurement or inflow forecasts. The sensitivity test of different rationing rates depending on the stored volume can analyze the time of response of each system. The change in recovery time, however, was not very significant.

The world's water resources face unsustainable pressure from population growth, changes in consumption patterns, pollution, and overexploitation. Water resources managers have developed holistic approaches such as IWRM (Integrated Water Resources Management) and, more recently, the WEEF (Water-Energy-Environment-Food) nexus to address the situation. However, their application in day-to-day water resources management is still challenging due to the of little knowledge, data, and tools. One area where that challenge needs practical solutions is reservoir operation. The current study aims to improve the reservoir module in the Soil and Water Assessment Tool (SWAT) so that operation rules that aim to meet various water, food, and electricity objectives can be simulated. The improved SWAT model is used to simulate the management of the Sélingué reservoir in Mali, West Africa. The reservoir operation was simulated under three different operation rules: 1) priority to monthly hydropower production (HPP) target (rule 1); 2) respect of predefined monthly target storage (rule 2); 3) priority to downstream environmental flow, irrigation, and municipal water demands (rule 3). Results show that when priority is given to the HPP target (rule 1), 98.3% of the electricity demand is met. At the same time, the dam can supply 81.72% of the water demand to maintain environmental flow and sustain irrigation and municipal water consumption. It also ensures water availability with an annual target storage deviation estimated at 1.8%. When rule 2 is implemented, a gap of 8.5% between electricity production and electricity demand is observed. Rule 2 also failed to sustain environmental flow and supply flow for irrigation and municipal consumption as a gap of 15.39% between the supply and the demand was observed. Similarly to rule 1, It ensures water availability with an annual target storage deviation estimated at 1.25%. When rule 3 is enforced (i.e., the priority is given to environmental flow, irrigation, and municipal water demands) the reservoir can maintain the environmental flow and maintain irrigation, and municipal water requirements with a gap of 17.7% between the supply and the demands. However, HPP production decreases with a gap of 12.56% between the electricity supply and demand. Its capacity to supply water in the long term is low as it has the highest target storage deviation with a value of 18%. These results indicate that rule 1 offers more guarantees considering the food and electricity security and environmental challenges. Note that the simulations are done assuming that these rules are systematically followed. In practice, decision-makers can deviate from a rule in exceptional circumstances to maximize benefits or avert unwanted consequences. Finally, a decision support system (DSS) was developed to assist decision-makers in selecting efficient reservoir operation policies for multipurpose reservoirs combining HPP and irrigation.

Diversas atividades humanas são fortemente dependentes do clima e da sua variabilidade, especialmente aquelas relacionadas ao uso da água. A operação integrada de reservatórios com múltiplos usos requer uma série de decisões que definem quanta água deve ser alocada, ao longo do tempo para cada um dos usos, e quais os volumes dos reservatórios a serem mantidos. O conhecimento antecipado das condições climáticas resulta de vital importância para os operadores de reservatórios, pois o insumo dos reservatórios é a vazão dos rios, que por sua vez é dependente de condições atmosféricas e hidrológicas em diferentes escalas de tempo e espaço. A pesquisa trata sobre três importantes elementos de subsídio à tomada de decisão na operação de reservatórios baseada na previsão de variáveis hidrológicas: (a) as previsões de vazão de curto prazo; (b) as previsões de precipitação de longo prazo e (c) as medidas de desempenho das previsões. O reservatório de Furnas, localizado na bacia do Rio Grande, em Minas Gerais, foi selecionado como estudo de caso devido, principalmente, à disponibilidade de previsões quantitativas de chuva e pela importância desse reservatório na região analisada. A previsão de curto prazo de vazão com base na precipitação foi estimada com um modelo empírico (rede neural artificial) e a previsão de precipitação foi obtida pelo modelo regional ETA. Uma metodologia de treinamento e validação da rede neural artificial foi desenvolvida utilizando previsões perfeitas de chuva (considerando a chuva observada como previsão) e utilizando o maior número de dados disponíveis, favorecendo a representatividade dos resultados obtidos. A metodologia empírica alcançou os desempenhos obtidos com um modelo hidrológico conceitual, mostrando-se menos sensitiva aos erros na previsão quantitativa de precipitação nessa bacia. Os resultados obtidos mostraram que as previsões de vazão utilizando modelos empíricos e conceituais e incorporando previsões quantitativas de precipitação são melhores que a metodologia utilizada pelo ONS no local de estudo. A redução dos erros de previsão relativos à metodologia empregada pelo ONS foi em torno de 20% quando usadas previsões quantitativas de precipitação definidas pelo modelo regional ETA e superiores a 50% quando usadas previsões perfeitas de precipitação. Embora essas últimas previsões nunca possam ser obtidas na prática, os resultados sugerem o quanto o incremento do desempenho das previsões quantitativas de chuva melhoraria as previsões de vazão. A previsão de precipitação de longo prazo para a bacia analisada foi também estimada com um modelo empírico de redes neurais artificiais e utilizando índices climáticos como variáveis de entrada. Nesse sentido, foram estimadas previsões de precipitação acumulada no período mais chuvoso (DJF) utilizando índices climáticos associados a fenômenos climáticos, como o El Niño - Oscilação Sul e a Oscilação Decadal do Pacífico, e a modos de variabilidade climática, como a Oscilação do Atlântico Norte e o Modo Anular do Hemisfério Sul. Apesar das redes neurais artificiais terem sido aplicadas em diversos problemas relacionados a hidrometeorologia, a aplicação dessas técnicas na previsão de precipitação de longo prazo é ainda rara. Os resultados obtidos nesse trabalho mostraram que consideráveis reduções dos erros da previsão relativos ao uso apenas da média climatológica como previsão podem ser obtidos com a metodologia utilizada. Foram obtidas reduções dos erros de, no mínimo 50%, e chegando até um valor próximo a 75% nos diferentes testes efetuados no estudo de caso. Uma medida de desempenho da previsão foi desenvolvida baseada no uso de tabelas de contingência e levando em conta a utilidade da previsão. Essa medida de desempenho foi calculada com base nos resultados do uso das previsões por um modelo de operação de reservatório, e não apenas na comparação de vazões previstas e observadas. Nos testes realizados durante essa pesquisa, ficou evidente que não existe uma relação unívoca entre qualidade das previsões e utilidade das previsões. No entanto, em função de comportamentos particulares das previsões, tendências foram encontradas, como por exemplo nos modelos cuja previsão apresenta apenas defasagem. Nesses modelos, a utilidade das previsões tende a crescer na medida que a qualidade das mesmas aumenta. Por fim, uma das grandes virtudes da medida de desempenho desenvolvida nesse trabalho foi sua capacidade de distinguir o desempenho de modelos que apresentaram a mesma qualidade. / Several human activities are strongly dependent on climate and its variability, especially those related to water use. The operation of multi-purpose reservoirs systems defines how much water should be allocated and the reservoir storage volumes to be maintained, over time. Knowing in advance the weather conditions helps the decision making process, as the major inputs to reservoirs are the streamflows, which are dependent on atmospheric and hydrological conditions at different time-space scales. This research deals with three important aspects towards the decision making process of multi-purpose reservoir operation based on forecast of hydrological variables: (a) short-term streamflow forecast, (b) long-range precipitation forecast and (c) performance measures. The Furnas reservoir on the Rio Grande basin was selected as the case study, primarily because of the availability of quantitative precipitation forecasts from the Brazilian Center for Weather Prediction and Climate Studies and due to its importance in the Brazilian hydropower generation system. Short-term streamflow forecasts were estimated by an empirical model (artificial neural network – ANN) and incorporating forecast of rainfall. Quantitative precipitation forecasts (QPFs), defined by the ETA regional model, were used as inputs to the ANN models. A methodology for training and validating the ANN models was developed using perfect precipitation forecasts (i.e., using the observed precipitation as if it was a forecast) and considering the largest number of available samples, in order to increase the representativeness of the results. The empirical methodology achieved the performance obtained with a conceptual hydrological model and seemed to be less sensitive to precipitation forecast error relative to the conceptual hydrological model. Although limited to one reservoir, the results obtained show that streamflow forecasting using empirical and conceptual models and incorporating QPFs performs better than the methodology used by ONS. Reduction in the forecast errors relative to the ONS method was about 20% when using QPFs provided by ETA model, and greater than 50% when using the perfect precipitation forecast. Although the latter can never be achieved in practice, these results suggest that improving QPFs would lead to better forecasts of reservoir inflows. Long-range precipitation forecast was also estimated by an empirical model based on artificial neural networks and using climate indices as input variables. The output variable is the summer (DJF) precipitation over the Furnas watershed. It was estimated using climate indices related to climatic phenomena such as El Niño - Southern Oscillation and the Pacific Decadal Oscillation and modes of climate variability, such as the North Atlantic Oscillation and the Southern Annular Mode. Despite of ANN has been applied in several problems of hydrometeorological areas, the application of such technique for long-range precipitation forecast is still rare. The results obtained demonstrate how the methodology for seasonal precipitation forecast based on ANN can be particularly helpful, with the use of available time series of climate indices. Reductions in the forecast errors achieved by using only the climatological mean as forecast were considerable, being at least of 50% and reaching values close to 75% in several tests. A performance measure based on the use of contingency tables was developed taking into account the utility of the forecast. This performance measure was calculated based on the results of the use of the forecasts by a reservoir operation model, and not only by comparing streamflow observed and forecast. The performed tests show that there is no unequivocal relationship between quality and utility of the forecasts. However, when the forecast has a particular behavior, trends were found in the relationship between utility and quality of the forecast, such as models that generate streamflow forecast with lags in comparison to the observed values. In these models, the utility of the forecasts tends to enhance as the quality increases. Finally, the ability to distinguish the performance of forecast models having similar quality was one of the main merits of the performance measure developed in this research.

Diversas atividades humanas são fortemente dependentes do clima e da sua variabilidade, especialmente aquelas relacionadas ao uso da água. A operação integrada de reservatórios com múltiplos usos requer uma série de decisões que definem quanta água deve ser alocada, ao longo do tempo para cada um dos usos, e quais os volumes dos reservatórios a serem mantidos. O conhecimento antecipado das condições climáticas resulta de vital importância para os operadores de reservatórios, pois o insumo dos reservatórios é a vazão dos rios, que por sua vez é dependente de condições atmosféricas e hidrológicas em diferentes escalas de tempo e espaço. A pesquisa trata sobre três importantes elementos de subsídio à tomada de decisão na operação de reservatórios baseada na previsão de variáveis hidrológicas: (a) as previsões de vazão de curto prazo; (b) as previsões de precipitação de longo prazo e (c) as medidas de desempenho das previsões. O reservatório de Furnas, localizado na bacia do Rio Grande, em Minas Gerais, foi selecionado como estudo de caso devido, principalmente, à disponibilidade de previsões quantitativas de chuva e pela importância desse reservatório na região analisada. A previsão de curto prazo de vazão com base na precipitação foi estimada com um modelo empírico (rede neural artificial) e a previsão de precipitação foi obtida pelo modelo regional ETA. Uma metodologia de treinamento e validação da rede neural artificial foi desenvolvida utilizando previsões perfeitas de chuva (considerando a chuva observada como previsão) e utilizando o maior número de dados disponíveis, favorecendo a representatividade dos resultados obtidos. A metodologia empírica alcançou os desempenhos obtidos com um modelo hidrológico conceitual, mostrando-se menos sensitiva aos erros na previsão quantitativa de precipitação nessa bacia. Os resultados obtidos mostraram que as previsões de vazão utilizando modelos empíricos e conceituais e incorporando previsões quantitativas de precipitação são melhores que a metodologia utilizada pelo ONS no local de estudo. A redução dos erros de previsão relativos à metodologia empregada pelo ONS foi em torno de 20% quando usadas previsões quantitativas de precipitação definidas pelo modelo regional ETA e superiores a 50% quando usadas previsões perfeitas de precipitação. Embora essas últimas previsões nunca possam ser obtidas na prática, os resultados sugerem o quanto o incremento do desempenho das previsões quantitativas de chuva melhoraria as previsões de vazão. A previsão de precipitação de longo prazo para a bacia analisada foi também estimada com um modelo empírico de redes neurais artificiais e utilizando índices climáticos como variáveis de entrada. Nesse sentido, foram estimadas previsões de precipitação acumulada no período mais chuvoso (DJF) utilizando índices climáticos associados a fenômenos climáticos, como o El Niño - Oscilação Sul e a Oscilação Decadal do Pacífico, e a modos de variabilidade climática, como a Oscilação do Atlântico Norte e o Modo Anular do Hemisfério Sul. Apesar das redes neurais artificiais terem sido aplicadas em diversos problemas relacionados a hidrometeorologia, a aplicação dessas técnicas na previsão de precipitação de longo prazo é ainda rara. Os resultados obtidos nesse trabalho mostraram que consideráveis reduções dos erros da previsão relativos ao uso apenas da média climatológica como previsão podem ser obtidos com a metodologia utilizada. Foram obtidas reduções dos erros de, no mínimo 50%, e chegando até um valor próximo a 75% nos diferentes testes efetuados no estudo de caso. Uma medida de desempenho da previsão foi desenvolvida baseada no uso de tabelas de contingência e levando em conta a utilidade da previsão. Essa medida de desempenho foi calculada com base nos resultados do uso das previsões por um modelo de operação de reservatório, e não apenas na comparação de vazões previstas e observadas. Nos testes realizados durante essa pesquisa, ficou evidente que não existe uma relação unívoca entre qualidade das previsões e utilidade das previsões. No entanto, em função de comportamentos particulares das previsões, tendências foram encontradas, como por exemplo nos modelos cuja previsão apresenta apenas defasagem. Nesses modelos, a utilidade das previsões tende a crescer na medida que a qualidade das mesmas aumenta. Por fim, uma das grandes virtudes da medida de desempenho desenvolvida nesse trabalho foi sua capacidade de distinguir o desempenho de modelos que apresentaram a mesma qualidade. / Several human activities are strongly dependent on climate and its variability, especially those related to water use. The operation of multi-purpose reservoirs systems defines how much water should be allocated and the reservoir storage volumes to be maintained, over time. Knowing in advance the weather conditions helps the decision making process, as the major inputs to reservoirs are the streamflows, which are dependent on atmospheric and hydrological conditions at different time-space scales. This research deals with three important aspects towards the decision making process of multi-purpose reservoir operation based on forecast of hydrological variables: (a) short-term streamflow forecast, (b) long-range precipitation forecast and (c) performance measures. The Furnas reservoir on the Rio Grande basin was selected as the case study, primarily because of the availability of quantitative precipitation forecasts from the Brazilian Center for Weather Prediction and Climate Studies and due to its importance in the Brazilian hydropower generation system. Short-term streamflow forecasts were estimated by an empirical model (artificial neural network – ANN) and incorporating forecast of rainfall. Quantitative precipitation forecasts (QPFs), defined by the ETA regional model, were used as inputs to the ANN models. A methodology for training and validating the ANN models was developed using perfect precipitation forecasts (i.e., using the observed precipitation as if it was a forecast) and considering the largest number of available samples, in order to increase the representativeness of the results. The empirical methodology achieved the performance obtained with a conceptual hydrological model and seemed to be less sensitive to precipitation forecast error relative to the conceptual hydrological model. Although limited to one reservoir, the results obtained show that streamflow forecasting using empirical and conceptual models and incorporating QPFs performs better than the methodology used by ONS. Reduction in the forecast errors relative to the ONS method was about 20% when using QPFs provided by ETA model, and greater than 50% when using the perfect precipitation forecast. Although the latter can never be achieved in practice, these results suggest that improving QPFs would lead to better forecasts of reservoir inflows. Long-range precipitation forecast was also estimated by an empirical model based on artificial neural networks and using climate indices as input variables. The output variable is the summer (DJF) precipitation over the Furnas watershed. It was estimated using climate indices related to climatic phenomena such as El Niño - Southern Oscillation and the Pacific Decadal Oscillation and modes of climate variability, such as the North Atlantic Oscillation and the Southern Annular Mode. Despite of ANN has been applied in several problems of hydrometeorological areas, the application of such technique for long-range precipitation forecast is still rare. The results obtained demonstrate how the methodology for seasonal precipitation forecast based on ANN can be particularly helpful, with the use of available time series of climate indices. Reductions in the forecast errors achieved by using only the climatological mean as forecast were considerable, being at least of 50% and reaching values close to 75% in several tests. A performance measure based on the use of contingency tables was developed taking into account the utility of the forecast. This performance measure was calculated based on the results of the use of the forecasts by a reservoir operation model, and not only by comparing streamflow observed and forecast. The performed tests show that there is no unequivocal relationship between quality and utility of the forecasts. However, when the forecast has a particular behavior, trends were found in the relationship between utility and quality of the forecast, such as models that generate streamflow forecast with lags in comparison to the observed values. In these models, the utility of the forecasts tends to enhance as the quality increases. Finally, the ability to distinguish the performance of forecast models having similar quality was one of the main merits of the performance measure developed in this research.

Diversas atividades humanas são fortemente dependentes do clima e da sua variabilidade, especialmente aquelas relacionadas ao uso da água. A operação integrada de reservatórios com múltiplos usos requer uma série de decisões que definem quanta água deve ser alocada, ao longo do tempo para cada um dos usos, e quais os volumes dos reservatórios a serem mantidos. O conhecimento antecipado das condições climáticas resulta de vital importância para os operadores de reservatórios, pois o insumo dos reservatórios é a vazão dos rios, que por sua vez é dependente de condições atmosféricas e hidrológicas em diferentes escalas de tempo e espaço. A pesquisa trata sobre três importantes elementos de subsídio à tomada de decisão na operação de reservatórios baseada na previsão de variáveis hidrológicas: (a) as previsões de vazão de curto prazo; (b) as previsões de precipitação de longo prazo e (c) as medidas de desempenho das previsões. O reservatório de Furnas, localizado na bacia do Rio Grande, em Minas Gerais, foi selecionado como estudo de caso devido, principalmente, à disponibilidade de previsões quantitativas de chuva e pela importância desse reservatório na região analisada. A previsão de curto prazo de vazão com base na precipitação foi estimada com um modelo empírico (rede neural artificial) e a previsão de precipitação foi obtida pelo modelo regional ETA. Uma metodologia de treinamento e validação da rede neural artificial foi desenvolvida utilizando previsões perfeitas de chuva (considerando a chuva observada como previsão) e utilizando o maior número de dados disponíveis, favorecendo a representatividade dos resultados obtidos. A metodologia empírica alcançou os desempenhos obtidos com um modelo hidrológico conceitual, mostrando-se menos sensitiva aos erros na previsão quantitativa de precipitação nessa bacia. Os resultados obtidos mostraram que as previsões de vazão utilizando modelos empíricos e conceituais e incorporando previsões quantitativas de precipitação são melhores que a metodologia utilizada pelo ONS no local de estudo. A redução dos erros de previsão relativos à metodologia empregada pelo ONS foi em torno de 20% quando usadas previsões quantitativas de precipitação definidas pelo modelo regional ETA e superiores a 50% quando usadas previsões perfeitas de precipitação. Embora essas últimas previsões nunca possam ser obtidas na prática, os resultados sugerem o quanto o incremento do desempenho das previsões quantitativas de chuva melhoraria as previsões de vazão. A previsão de precipitação de longo prazo para a bacia analisada foi também estimada com um modelo empírico de redes neurais artificiais e utilizando índices climáticos como variáveis de entrada. Nesse sentido, foram estimadas previsões de precipitação acumulada no período mais chuvoso (DJF) utilizando índices climáticos associados a fenômenos climáticos, como o El Niño - Oscilação Sul e a Oscilação Decadal do Pacífico, e a modos de variabilidade climática, como a Oscilação do Atlântico Norte e o Modo Anular do Hemisfério Sul. Apesar das redes neurais artificiais terem sido aplicadas em diversos problemas relacionados a hidrometeorologia, a aplicação dessas técnicas na previsão de precipitação de longo prazo é ainda rara. Os resultados obtidos nesse trabalho mostraram que consideráveis reduções dos erros da previsão relativos ao uso apenas da média climatológica como previsão podem ser obtidos com a metodologia utilizada. Foram obtidas reduções dos erros de, no mínimo 50%, e chegando até um valor próximo a 75% nos diferentes testes efetuados no estudo de caso. Uma medida de desempenho da previsão foi desenvolvida baseada no uso de tabelas de contingência e levando em conta a utilidade da previsão. Essa medida de desempenho foi calculada com base nos resultados do uso das previsões por um modelo de operação de reservatório, e não apenas na comparação de vazões previstas e observadas. Nos testes realizados durante essa pesquisa, ficou evidente que não existe uma relação unívoca entre qualidade das previsões e utilidade das previsões. No entanto, em função de comportamentos particulares das previsões, tendências foram encontradas, como por exemplo nos modelos cuja previsão apresenta apenas defasagem. Nesses modelos, a utilidade das previsões tende a crescer na medida que a qualidade das mesmas aumenta. Por fim, uma das grandes virtudes da medida de desempenho desenvolvida nesse trabalho foi sua capacidade de distinguir o desempenho de modelos que apresentaram a mesma qualidade. / Several human activities are strongly dependent on climate and its variability, especially those related to water use. The operation of multi-purpose reservoirs systems defines how much water should be allocated and the reservoir storage volumes to be maintained, over time. Knowing in advance the weather conditions helps the decision making process, as the major inputs to reservoirs are the streamflows, which are dependent on atmospheric and hydrological conditions at different time-space scales. This research deals with three important aspects towards the decision making process of multi-purpose reservoir operation based on forecast of hydrological variables: (a) short-term streamflow forecast, (b) long-range precipitation forecast and (c) performance measures. The Furnas reservoir on the Rio Grande basin was selected as the case study, primarily because of the availability of quantitative precipitation forecasts from the Brazilian Center for Weather Prediction and Climate Studies and due to its importance in the Brazilian hydropower generation system. Short-term streamflow forecasts were estimated by an empirical model (artificial neural network – ANN) and incorporating forecast of rainfall. Quantitative precipitation forecasts (QPFs), defined by the ETA regional model, were used as inputs to the ANN models. A methodology for training and validating the ANN models was developed using perfect precipitation forecasts (i.e., using the observed precipitation as if it was a forecast) and considering the largest number of available samples, in order to increase the representativeness of the results. The empirical methodology achieved the performance obtained with a conceptual hydrological model and seemed to be less sensitive to precipitation forecast error relative to the conceptual hydrological model. Although limited to one reservoir, the results obtained show that streamflow forecasting using empirical and conceptual models and incorporating QPFs performs better than the methodology used by ONS. Reduction in the forecast errors relative to the ONS method was about 20% when using QPFs provided by ETA model, and greater than 50% when using the perfect precipitation forecast. Although the latter can never be achieved in practice, these results suggest that improving QPFs would lead to better forecasts of reservoir inflows. Long-range precipitation forecast was also estimated by an empirical model based on artificial neural networks and using climate indices as input variables. The output variable is the summer (DJF) precipitation over the Furnas watershed. It was estimated using climate indices related to climatic phenomena such as El Niño - Southern Oscillation and the Pacific Decadal Oscillation and modes of climate variability, such as the North Atlantic Oscillation and the Southern Annular Mode. Despite of ANN has been applied in several problems of hydrometeorological areas, the application of such technique for long-range precipitation forecast is still rare. The results obtained demonstrate how the methodology for seasonal precipitation forecast based on ANN can be particularly helpful, with the use of available time series of climate indices. Reductions in the forecast errors achieved by using only the climatological mean as forecast were considerable, being at least of 50% and reaching values close to 75% in several tests. A performance measure based on the use of contingency tables was developed taking into account the utility of the forecast. This performance measure was calculated based on the results of the use of the forecasts by a reservoir operation model, and not only by comparing streamflow observed and forecast. The performed tests show that there is no unequivocal relationship between quality and utility of the forecasts. However, when the forecast has a particular behavior, trends were found in the relationship between utility and quality of the forecast, such as models that generate streamflow forecast with lags in comparison to the observed values. In these models, the utility of the forecasts tends to enhance as the quality increases. Finally, the ability to distinguish the performance of forecast models having similar quality was one of the main merits of the performance measure developed in this research.

L’énergie nucléaire, qui est une technologie de génération largement utilisée dans des systèmes électriques, est caractérisée par des coûts fixes élevés et des coûts variables bas. Pour amortir ses coûts fixes, le nucléaire est préférentiellement utilisé pour une opération en base inflexible, c’est-à-dire opérer à un niveau constant pour répondre à la partie non variable de la demande d’un système électrique et produire au maximum de sa propre capacité. En raison de cette spécificité, l’insertion de la production nucléaire dans les marchés concurrentiels d’électricité n’a pas été profondément étudiée jusqu’à présent. Par conséquent, même dans des marchés concurrentiels, la question de la gestion optimale d’un parc de production nucléaire n’a pas été soulevée parce que la production nucléaire est censée fonctionner en continu (pour couvrir la demande de base). Cependant, il y a des cas ou` la gestion de la production nucléaire semble plus complexe que ne le suggère cette vision simplifiée. En règle générale, lorsque la proportion de l’énergie nucléaire dans un parc de production est élevée, la production nucléaire doit s’adapter aux variations de la demande. Cela soulève la question de la façon optimale de gérer cette technologie de production dans ce contexte. Comme cette question n’a pas été étudiée jusqu’à présent, il est nécessaire de proposer un cadre théorique qui permet une analyse des situations comme celle de la France, avec un marché concurrentiel et où le nucléaire représente 80% de la production, c’est-à-dire beaucoup plus que ce qui serait nécessaire pour couvrir la demande de base. Nous nous plaçons dans un horizon à moyen terme de la gestion (1 à 3 ans) pour tenir compte de la variation saisonnière de la demande. A moyen terme, le gestionnaire d’un parc nucléaire très large (comme le parc français) doit ajuster sa production selon les variations saisonnières de la demande. Dans ce cadre, le stock de combustible nucléaire peut être analysé comme un réservoir puisque les centrales nucléaires s’arrêtent périodiquement (tous les 12 ou 18 mois) pour recharger leur combustible. La gestion de ce réservoir permet de profils différents d’usages de combustible nucléaire au cours des différentes saisons de l’année. Ainsi, nous nous pencherons sur cette question comme une analyse économique rationnelle de l’opération d’un “réservoir” de combustible nucléaire. Nous allons ensuite l’analyser dans un cadre général déterministe dynamique avec deux types de production : nucléaire et thermique non-nucléaire. Nous étudions la gestion optimale de la production dans un marché parfaitement concurrentiel. Ensuite, nous établissons un modèle numérique (basé sur les données du marché français) où les centrales nucléaires ne sont pas opérées à production constante, mais dans un cadre de placement flexible (comme le parc nucléaire français). […] / Nuclear power as a generation technology that is widely used in electricity production systems is characterized by high fixed costs and low variable costs. To amortize its fixed costs, nuclear is preferentially used for inflexible baseload operation, i.e. operate at a constant level to meet the non variable part of electricity demand of a system and produce at its maximum capacity. Because of this specificity, the insertion of nuclear production in competitive electricity markets has not been deeply studied so far. Therefore, even in competitive markets, the question of the optimal management of a nuclear generation set has not been raised because nuclear production is supposed to operate continuously (to cover baseline demand). However, there are cases where the management of nuclear generation seems more complex than suggested by this simplified view. Typically, when the proportion of nuclear energy in a production set is high, the nuclear generation output has to adjust to the variations in demand. This raises the question of the optimal way to manage this production technology in that kind of setting. As this question has not been studied so far, there is a need for a theoretical framework that enables an analysis of situations like the French one, with a competitive market and where nuclear represents 80% of generation, i.e. much more that what would be necessary to cover the baseload demand. We place ourselves in a medium-term horizon of the management (1 to 3 years) to take into account the seasonal variation of the demand level. In the medium-term, the manager of a large nuclear set (like the French set) has to set its seasonal variation of output according to the demand level. Since nuclear units have to stop periodically (from 12 to 18 months) to reload their fuel, we can analyze the nuclear fuel as a stock behaving like a reservoir. The operation of the reservoir allows different profiles of nuclear fuel use during the different demand seasons of the year. Thus, we will look at this question as a rational economic analysis of the operation of a nuclear fuel “reservoir”. We then analyze it within a general deterministic dynamic framework with two types of generation: nuclear and thermal non-nuclear. We study the optimal management of the production in a perfectly competitive market. Then, we establish a numerical model (based on data from the French market) with nuclear plants being not operated strictly as base load power plants but within a flexible dispatch frame (like the French nuclear set). [...]

Optimal operation of reservoir systems is necessary for better utilizing the limited water resources and to justify the high capital investments associated with reservoir projects. However, finding optimal policies for real-life problems of reservoir systems operation (RSO) is a challenging task as the available analytical methods can not handle the arbitrary functions of the problem and almost all methods employed are numerical or iterative type that are computer dependent. Since the computer resources in terms of memory and CPU time are limited, a limit exists for the size of the problem, in terms of arithmetic and memory involved, that can be handled. This limit is approached quickly as the dimension and the nonlinearity of the problem increases. In encountering the complex aspects of the problem all the traditionally employed methods have their own drawbacks. Linear programming (LP), though very efficient in dealing with linear functions, can not handle nonlinear functions which is the case mostly in real-life problems. Attempting to approximate nonlinear functions to linear ones results in the problem size growing enormously. Dynamic programming (DP), though suitable for most of the RSO problems, requires exponentially increasing computer resources as the dimension of the problem increases and at present many high dimensional real-life problems can not be solved using DP. Nonlinear programming (NLP) methods are not known to be efficient in RSO problems due to slow rate of convergence and inability to handle stochastic problems. Simulation methods can, practically, explore only a small portion of the search region. Many simplifications in formulations and adoption of approximate methods in literature still fall short in addressing the most critical aspects, namely multidimensionality, stochasticity, and additional complexity in conjunctive operation, of the problem. As the problem complexity increases and the possibility of arriving at the solution recedes, a near optimal solution with the best use of computational resources can be very valuable. In this context, genetic algorithms (GA) can be a promising technique which is believed to have an advantage in terms of efficient use of computer resources. GA is a random search method which find, in general, near optimal solutions using evolutionary mechanism of natural selection and natural genetics. When a pool of feasible solutions, represented in a coded form, are given fitness according to a objective function and explored by genetic operators for obtaining new pools of solutions, then the ensuing trajectories of solutions come closer and closer to the optimal solution which has the greatest fitness associated with it. GA can be applied to arbitrary functions and is not excessively sensitive to the dimension of the problem. Though in general GA finds only the near optimal solutions trapping in local optima is not a serious problem due to global look and random search. Since GA is not fully explored for RSO problems two such problems are selected here to study the usefulness and efficiency of GA in obtaining near optimal solutions. One problem is conjunctive operation of a system consisting of a surface reservoir and an aquifer, taken from the literature for which deterministic and stochastic models are solved. Another problem is real-time operation of a multipurpose reservoir, operated for irrigation (primary purpose) and hydropower production, which is in the form of a case study. The conjunctive operation problem consists of determining optimal policy for a combined system of a surface reservoir and an aquifer. The surface reservoir releases water to an exclusive area for irrigation and to a recharge facility from which it reaches the aquifer in the following period. Another exclusive area is irrigated by water pumped from the aquifer. The objective is to maximize the total benefit from the two irrigated areas. The inflow to the surface reservoir is treated as constant in deterministic model and taken at 6 different classes in stochastic model. The hydrological interactions between aquifer and reservoir are described using a lumped parameter model in which the average aquifer water table is arrived at based on the quantity of water in the aquifer, and local drawdown in pumping well is neglected. In order to evaluate the GA solution both deterministic and stochastic models are solved using DP and stochastic DP (SDP) techniques respectively. In the deterministic model, steady state (SS) cyclic (repetitive) solution is identified in DP as well as in GA. It is shown that the benefit from GA solution converges to as near as 95% of the benefit from exact DP solution at a highly discounted CPU time. In the stochastic model, the steady state solution obtained with SDP consists of converged first stage decisions, which took a 8-stage horizon, for any combination of components of the system state. The GA solution is obtained after simplifying the model to reduce the number of decision variables. Unlike SDP policy which gives decisions considering the state of the system in terms of storages, at reservoir, aquifer, and recharge facility, and previous inflow at the beginning of that period, GA gives decisions for each period of the horizon considering only the past inflow state of the period. In arriving at these decisions the effect of neglected state information is approximately reflected in the decisions by the process of refinement of the decisions, to conform to feasibility of storages in reservoir and aquifer, carried out in a simplified simulation process. Moreover, the validity of the solution is confirmed by simulating the operation with all possible inflow sequences for which the 8-stages benefit converged up to 90 % of the optimum. However, since 8 stages are required for convergence to SS, a 16-stage process is required for GA method in which the first 8 stages policy is valid. Results show that GA convergence to the optimum is satisfactory, justifying the approximations, with significant savings in CPU time. For real-time operation of a multipurpose reservoir, a rule curve (RC) based monthly operation is formulated and applied on a real-life problem involving releases for irrigation as well as power production. The RC operation is based on the target storages that have to be maintained, at each season of the year, in the reservoir during normal hydrological conditions. Exceptions to target storages are allowed when the demands have to be met or for conserving water during the periods of high inflows. The reservoir in the case study supplies water to irrigation fields through two canals where a set of turbines each at the canal heads generate hydropower. A third set of turbines operate on the river bed with the water let out downstream from the dam. The problem consists of determining the the RC target storages that facilitate maximum power production while meeting the irrigation demands up to a given reliability level. The RC target storages are considered at three different levels, corresponding to dry, normal, and wet conditions, according to the system state in terms of actual (beginning of period) storage of the reservoir. That is, if the actual beginning storage of the reservoir is less than some coefficient, dry-coe, times the normal target storage the target for the end of the period storage is taken at the dry storage target (of the three sets of storages). Similarly the wet level is taken for the end of the period target if the actual beginning storage is greater than some coefficient, wet-coe, times the normal storage. For other conditions the target is the normal storage level. The dry-coe and wet-coe parameters are obtained by trial and error analysis working on a small sequence of inflows. The three sets of targets are obtained from optimization over a 1000 year generated inflow sequence. With deterministic DP solutions, for small sequences of inflows, the optimization capability of GA-RC approach, in terms of objective function convergence, and generalization or robustness capability of GA-RC approach, for which the GA-RC benefit is obtained by simulating the reservoir operation using the previously obtained GA-RC solution, are evaluated. In both the cases GA-RC approach proves to be promising. Finally a 15 year real-time simulation of the reservoir is carried out using historical inflows and demands and the comparison with the historical operation shows significant improvement in benefit, i.e. power produced, without compromising irrigation demands throughout the simulation period.

From the Proceedings of the 1977 Meetings of the Arizona Section - American Water Resources Assn. and the Hydrology Section - Arizona Academy of Science - April 15-16, 1977, Las Vegas, Nevada / A dual purpose reservoir control problem can logically be modelled as a game against nature. The first purpose of the reservoir is flood control under uncertain inflow, which corresponds to short -range operation (SRO); the second purpose, which the present model imbeds into the first one, is water supply after the flood has receded, and corresponds to long-range operation (LRO). The reservoir manager makes release decisions based on his SRO risk. The trade-offs involved in his decision are described by a utility function, which is constructed within the framework of Keeney's multiattribute utility theory. The underlying assumptions appear to be quite natural for the reservoir control problem. To test the model, an experiment assessing the utility criterion of individuals has been performed; the results tend to confirm the plausibility of the approach. In particular, most individuals appear to have a risk-averse attitude for small floods and a risk-taking attitude for large ones.