Abstract
A methodology to forecast power demand of oil and gas installations which uses publicly production available data, parametric models and data-driven Gaussian regression methods is presented. The methodology also captures the expected fuel gas consumption and energy ratio. The proposed methodology is tested on the Brage field on the Norwegian Continental Shelf. It is shown that the general oil and water production behaviour as well as fuel gas consumption trends can be predicted. However, the forecast inherits a significant uncertainty due to the publicly available dataset lacking metadata and a complete description of the energy sinks. © 2024 Elsevier B.V. Author keywords Forecast; Gaussian Process Regression; Oil and Gas; Power demand