In the early 21st century, oil and gas production in the U.S. was conjectured to be in terminal-irreversible decline. But, thanks to the advancement of hydraulic fracturing technologies over the last decade, operators are now able to produce two-thirds of U.S. oil and gas output from almost impermeable shale formations. Despite the enormous success of the ‘shale revolution’, there are still debates about how long shale production will last and if there will be enough to subsidize a meaningful transition to ‘greener’ power sources. Most official pronouncements of shale oil and gas reserves are based on purely empirical curve-fitting approaches or geological volumetric calculations that tend to largely overestimate the actual reserves. As an alternative to these industry-standard forecasting methods, we propose a more reliable, ‘transparent’, physics-guided and data-driven approach to estimating future production rates of oil and gas in shales. Our physics-based scaling method captures all essential physics of hydrocarbon production and hydrofracture geometry, yet it is as simple as the industry-favored Decline Curve Analysis (DCA), so that most engineers can adopt it. We also demonstrate that our method is as accurate as other analytical methods and has the same predictive power as commercial reservoir simulators but with less data required and significantly faster computational time. To capture the uncertainties of play-wide production, we combine physical scaling with the Generalized Extreme Value (GEV) statistics. So far, we have implemented this method to nearly half a million wells from all major U.S. shale plays. Since the results of our analyses are not subject to bias, policy-makers ought not to assume that the shale production boom will last for centuries.

Orientadores: Denis Jose Schiozer, Celio Maschio / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Mecanica e Instituto de Geociencias / Made available in DSpace on 2018-08-12T07:35:35Z (GMT). No. of bitstreams: 1 Becerra_GustavoGabriel_M.pdf: 16760750 bytes, checksum: 0609c24d13d46b9121f71356ce9d42a1 (MD5) Previous issue date: 2007 / Resumo: A escassez de informações de qualidade introduz risco ao processo de previsão da produção de petróleo tornando imprescindível o ajuste de histórico de produção, que é a calibração do modelo a partir da resposta produtiva registrada. O ajuste é um problema inverso, em que diferentes combinações dos valores dos parâmetros do reservatório podem conduzir a respostas aceitáveis, especialmente quando o grau de incerteza desses parâmetros é elevado. A integração do ajuste de histórico com a análise probabilística dos cenários representativos conduz à obtenção de uma metodologia para detecção dos modelos calibrados dentro de uma faixa de aceitaçãodefinida. O tratamento de atributos interdependentes de influência global e local e o avanço por etapas são necessários. Desta forma, o objetivo deste trabalho é apresentar uma metodologia que integra a análise de incertezas com o ajuste de histórico em modelos de reservatórios complexos. Este procedimento auxilia a detectar os atributos incertos críticos e sua possível variação com o intuito de estimar a faixa representativa das reservas a desenvolver. Não é alvo obter o melhor ajuste determinístico, mas refletir como o histórico possibilita uma mitigação das incertezas. Assim, a meta é usar modelos mais complexos e aprimorar a metodologia iniciada por Moura Filho (2006), desenvolvida para um modelo teórico simples. São utilizados dois casos de estudo de complexidade similar. Um deles referente ao reservatório do Campo de Namorado, utilizado para verificar e validar, em nível global, a aplicação da metodologia. Na etapa de aplicação, é usado um modelo sintético construído a partir de dados de afloramentos reais no Brasil e compreendendo informações de campos análogos com sistemas turbidíticos depositados em águas profundas. Os métodos aplicados, mediante a redefinição das probabilidades associadas e níveis dos atributos incertos, permitem: (1) reduzir a faixa de ajustes possíveis e obter modelos mais confiáveis; (2) identificar e condicionar à incerteza presente em função dos dados registrados; (3) diminuir os intervalos de incerteza dos parâmetros críticos identificados; (4) demarcar os limites seguros do desempenho futuro do reservatório. A conseqüência é um aumento da confiança no uso da simulação como ferramenta auxiliar do processo decisório. Além disso, procura-se fornecer à equipe multidisciplinar uma metodologia para reduzir o tempo empregado no gerenciamento de múltiplos atributos incertos na etapa de ajuste do modelo. / Abstract: The lack of reliable data or with high degree of uncertainty yields risk to the process of production prediction making the history matching, the model calibration from the registered field production indispensable. History matching is an inverse problem and, in general, different combinations of reservoir attributes can lead acceptable solutions, especially whit high degree of uncertainty of these attributes. The integration of history matching with a probabilistic analysis of representative models yields a way to detect matched models inside an acceptance interval, providing more efficient framework for predictions. It is necessary to consider dependences between global and local attributes. The scope of this work is to present a methodology that integrates the uncertainty analysis with the history matching process in complex models. This procedure helps to detect critical subsurface attributes and their possible variation, in order to estimate a representative range of the additional reserves to be developed. . It is not an objective to obtain the best deterministic model, but to mitigate uncertainties by using observed data. The objective is to improve the methodology initiated by Moura Filho (2006), applied to a simple model. The methodology presented in this work is applied in two study cases with similar complexity. Firstly, the methodology is verified and validated, on global scale, in Namorado Field. Then, at the application stage, it is chosen a synthetic reservoir model made from real outcrop data of Brazil and involving information from analog fields with turbiditic systems deposited in deep waters. The methodology allows the redefinition of the probability and levels of the dynamic and static attributes in order: (1) to reduce the group of possible history matching obtaining more realistic models; (2) to identify the existent uncertainty as a function of observed data; (3) to decrease the uncertainty range of critical reservoir parameters; (4) to increase the confidence in production forecast. One contribution of this work is to present a quantitative approach to increase the reliability on the use of reservoir simulation as an auxiliary tool in decision processes. Another purpose of this work is to provide a procedure to reduce the consumed time to handle multiples uncertainty attributes during the history matching. / Mestrado / Reservatórios e Gestão / Mestre em Ciências e Engenharia de Petróleo

[pt] A estimativa de produção de petróleo é um dos parâmetros essenciais para mensurar a economicidade de um campo e, para tanto, existem várias técnicas convencionais na área da engenharia de petróleo para predizer esse cálculo. Essas técnicas abrangem desde modelos analíticos simplificados até simulações numéricas mais complexas. Este trabalho propõem o uso de Redes Neurais Artificias (RNA) para prever uma curva de produção de óleo que mais se aproxime da obtida por um simulador numérico. A metodologia consiste na utilização da rede neural do tipo feedforward para a previsão da vazão inicial e da curva de produção ao longo de dez anos para um poço produtor de óleo. Essa metodologia tem aplicação prática na área da exploração, visto que, nessa fase, ainda há muita incerteza sobre a acumulação de petróleo e, portanto, os modelos de reservatório tendem a não ser complexos. Os resultados foram obtidos a partir do treinamento de RNAs com dados coletados do simulador numérico IMEX, cujas saídas foram posteriormente comparadas com os dados originais da simulação numérica. Foi possível obter uma precisão de 97 por cento na estimativa da vazão inicial do poço produtor de óleo. A previsão da curva de produção apresentou um erro percentual médio absoluto inferior a 10 por cento nos dois primeiros anos. Apesar dos valores de erro terem crescido ao longo dos últimos anos, eles são menores quando comparados com a metodologia de declínio exponencial e com a regressão linear múltipla. / [en] Production forecasting is one of the essential parameters to measure the economics of an oil field. There are several conventional techniques in petroleum engineering to estimate the production curve. They range from simplified analytical models to complex numerical simulations. This study proposes the use of Artificial Neural Networks (ANN) to predict an oil production curve that approximates to a numerical simulator curve. The methodology consists of using a feedforward neural network to predict the initial flow and the production forecast over ten years of an oil well. This methodology has practical application in the exploration area, since, at this stage, there is still much uncertainty about the oil accumulation, so the reservoir models tend not to be complex. The results were obtained from the ANN training with data collected from the numerical simulator IMEX, whose outputs were later compared with the original data of the numerical simulation. It was possible to get an estimate for the oil initial flow forecast with an accuracy of 97 percent. The production forecast had a mean absolute percentage error of less than 10 percent in the first two years. Despite the increasing error values over the years, they are smaller when compared to those obtained from the exponential decline and multiple linear regression.

The automation of industrial processes has been one of the greatest innovations in the industrial sector. It allows faster and accurate operations of production processes while producing more outputs than old manual production techniques. In the beverage industry, this innovation was also well embraced, especially to improve its bottling processes. However it has been proven that a continuous optimization of automation techniques using advanced and current trend of automation is the only way industrial companies will survive in a very competitive market. This becomes more challenging for small to medium scale enterprises (SMEs) which are not always keen in adopting new technologies by fear of overspending their little revenues. By doing so, SMEs are exposing themselves to limited growth and vulnerable lifecycle in this fast growing automation world. The main contribution of this study was to develop practical and affordable applications that will optimize the bottling process of a SME beverage plant by combining its existing production resources to basic principles of the current trend of automation, Industry 4.0 (I40). This research enabled the small beverage industry to achieve higher production rate, better delivery time and easy access of plant information through production forecast using linear regression, predictive maintenance using speed vibration sensor and decentralization of production monitoring via cloud applications. The existing plant Siemens S7-1200 programmable logic controller (PLC) and ZENON supervisory control and data acquisition (SCADA) system were used to program the optimized process with very few additional resources. This study also opened doors for automation in SMEs, in general, to use I40 in their production processes with available means and limited cost. / School of Computing / M.Tech (Engineering, Electrical)